Evidence · AFNI

3dAutomask

3dAutomask makes a brain mask of an EPI series or a skull-stripped image from its intensities alone. LARMORX has two ports of it, both in larmorx itself, and both give AFNI 26.3.05’s masks bit for bit in every compared case, fMRIPrep’s own calls included. The replica’s record has 125 cases. The clean room’s has 130; in 14 of them, fMRIPrep’s calls on images AFNI wrote, its header differs from AFNI’s where AFNI reads the geometry from its own header extension, in 7 significant digits (AFNI-10); its masks are AFNI’s.

Ports
  • replica larmorx Apache-2.0, with CC BY 4.0's attribution for the Medical College of Wisconsin's code: translated from AFNI's source; in pip install larmorx since 2026-10-29, until then the separate GPL package larmorx-gpl
  • clean room larmorx Apache-2.0: written from specs/3dAutomask.md
Status
replica validated; clean room validated (marked before the statuses' definition)
Records
the replica's: standard comparison (AFNI 26.3.05) · the clean room's: standard comparison (AFNI 26.3.05) · findings afni-masks-clean-room
Reference
3dAutomask · API page afni-automask

What is compared

3dAutomask makes a brain mask from the intensities alone. It averages the absolute values of a series’ volumes, finds a clip level from their histogram (gradually, octant by octant, unless -nograd), keeps the largest connected cluster above it, peels and unpeels it to cut off thin protrusions, and fills its holes. fMRIPrep calls it once per BOLD run, on its unifized BOLD reference, with -dilate 1 and -apply_prefix: the second skull strip of its BOLD reference workflow.

The two ports run beside AFNI 26.3.05’s 3dAutomask and AFNI 25.2.09’s on the same files with the same arguments, both AFNI releases the repository’s builds, and beside fMRIPrep’s own build of AFNI in its image.

bash
larmorx afni 3dAutomask -apply_prefix uni_xform_masked.nii.gz -dilate 1 -prefix uni_xform_mask.nii.gz uni_xform.nii.gz
python
import larmorx as lx

r = lx.afni.automask("uni_xform.nii.gz", dilate=1, apply=True)
r.mask, r.masked

The call above leaves implementation at "auto". The API page maps every option. How to read what follows: how to read the evidence.

Its two ports

3dAutomask’s results depend on AFNI library routines copyrighted by the Medical College of Wisconsin, GPL-2 until AFNI 25.2.09, so LARMORX has two ports of it. Since 2026-10-29 both are part of LARMORX itself (pip install larmorx). AFNI 26.3.05 releases the Medical College of Wisconsin’s code, which these programs’ results depend on, under CC BY 4.0. So the replica moved from the separate GPL package larmorx-gpl into LARMORX, under the Apache License with CC BY 4.0’s attribution (NOTICE, licensing), and runs in the same process. The replica is translated from AFNI’s source. The clean room is written without it, from a description of what the program does (specs/3dAutomask.md).

implementation="auto" | "replica" | "clean-room" chooses between them ("original" in 0.1.0rc1). On the command line, larmorx --implementation clean-room afni 3dAutomask ... does the same from the next release; in 0.1.0rc1 it is the environment variable LARMORX_IMPLEMENTATION. auto takes the replica (before 2026-10-29 only where larmorx-gpl was installed, the clean room otherwise). The examples below were computed with the replica as the separate program it was then; its records, made again in the same process on 2026-10-29, agree case for case.

The clean room decides each pass of its mask operations voxel by voxel in parallel, finds clusters by union-find over runs of voxels, and replaces two of AFNI’s repeated sweeps by breadth-first searches with the same result. Its result does not depend on the thread count.

The verdicts at a glance

Agreement, replica against the original: AFNI 3dAutomask 26.3.05 (standard comparison, 125 cases of the record), by category and verdict
options37
data types13
defects of AFNI7
dimensions9
errors19
expected divergences1
global options8
environment variables5
fMRIPrep's calls22
real images3
random testing1

Select a segment to list its cases below.

Agreement, clean room against the original: AFNI 3dAutomask 26.3.05 (standard comparison, 130 cases of the record), by category and verdict
options37
data types13
defects of AFNI7
dimensions9
errors19
expected divergences1
global options8
environment variables5
fMRIPrep's calls22
real images3
random testing6

Select a segment to list its cases below.

The clean room’s record runs 130 cases against AFNI 26.3.05, the replica’s 125. Since 2026-10-29 both ports follow AFNI 26.3.05, AFNI’s newest release, and their records run against it, with AFNI 25.2.09 run beside it on every case (AFNI’s own step): the replica since the version update, the clean room since its alignment the same day. Between the two releases these programs changed in exactly two places, both corrections LARMORX already made by default: header slice code 2 read as NIfTI defines it (AFNI-03), and 3dTshift’s -no_detrend de-meaning every voxel (AFNI-05). In reproduce both ports now give 26.3.05’s result there and name neither. The examples below show both versions’ outputs: on fMRIPrep’s calls they are the same bytes.

The replica gives AFNI’s bytes, data and NIfTI header, in every case that writes an output, and AFNI’s exit status and messages in every case. So does the clean room, but in 14 cases: fMRIPrep’s calls on its 3dUnifize outputs, which carry AFNI’s own header extension, where the header differs in 7 significant digits (AFNI-10, a deliberate correction); its masks are AFNI’s. Where AFNI crashes on a missing option value (AFNI-40), both ports stop with an error instead: the expected divergences. Each case is also run in fix, the default, where LARMORX corrects AFNI’s deterministic defects and names them.

In fMRIPrep: the BOLD reference’s second skull strip

fMRIPrep’s own call on one of its runs, given to AFNI, to fMRIPrep’s build of AFNI, to the replica and to the clean room (at 1 and 6 threads). Each mask is drawn over the image it was made from; choose Masked input above the panels for -apply_prefix’s output, and the difference panels for what separates two runs.

Worked example

fMRIPrep's second skull strip of a real BOLD reference

What this shows: On fMRIPrep’s call the replica and the clean room both give AFNI’s mask, every voxel, in both releases and in fMRIPrep’s build; the clean room on 1 and 6 threads. No correction by default changes this call.

The 3dAutomask node fMRIPrep 25.2.6 ran on the unifized BOLD reference of OpenNeuro ds005040 sub-002 (task clip, run 1): its arguments given to AFNI 25.2.09, to AFNI 26.3.05 (both ports’ reference since 2026-10-29), to fMRIPrep’s own build of AFNI in its image, to the replica and to LARMORX’s clean room (compared with AFNI 25.2.09 and 26.3.05, its record’s reference), in fix and in reproduce. The masks are drawn over the input; choose Masked input for -apply_prefix’s output.

Data
OpenNeuro ds005040, sub-002 (CC0-1.0): the BOLD reference fMRIPrep computed, N4-corrected, skull-stripped once and unifized (3dUnifize) by fMRIPrep; shown around the masks, dilated by 2 voxels
Node
fMRIPrep 25.2.6, bold_fit_wf/enhance_and_skullstrip_bold_wf/skullstrip_second_pass
Command
3dAutomask -apply_prefix uni_xform_masked.nii.gz -dilate 1 -prefix uni_xform_mask.nii.gz …/uni_xform.nii.gz
Programs
AFNI_25.2.09, gcc (Ubuntu 11.4.0-1ubuntu1~22.04.3) 11.4.0 (the repository’s build); AFNI AFNI_25.2.09 (Aug 24 2025) [64-bit] in fMRIPrep’s image; larmorx-gpl 0.0.1 (the replica), from LARMORX at commit 71f5223b7836; LARMORX 0.0.1 (the clean room) at commit 6f0b06b96112
Mask
49,308 of 163,840 voxels; the replica’s the same, voxel for voxel; the clean room’s the same, voxel for voxel
fMRIPrep's run
the mask its node wrote has the same bytes as its AFNI’s run here
Grid
64 × 64 × 40 voxels of 3 × 3 × 3.3 mm; shown: 54 × 63 × 40 around the brain
AFNI versions
AFNI 26.3.05 (both ports’ reference since 2026-10-29) on the same arguments as AFNI 25.2.09: the same bytes in every output of fMRIPrep’s call. What changed between the two releases in these programs (header slice code 2, AFNI-03; -no_detrend, AFNI-05; both corrected by LARMORX before AFNI fixed them) does not meet this call.
Inputds005040 sub-002 BOLD reference, unifized3dAutomask's input in fMRIPrep's BOLD reference workflow (float32)
OriginalAFNI 3dAutomask 25.2.09the repository's build
Replicareplicalarmorx, Apache-2.0 with CC BY 4.0's attribution · fix (the default) · against AFNI 3dAutomask 26.3.05, which it follows
Clean roomclean roomlarmorx, Apache-2.0 · fix (the default), 1 thread · against AFNI 3dAutomask 26.3.05, which it follows

Bit-identical163,840 of 163,840 voxels equal

the original against itself: fMRIPrep's build and the repository'sAFNI 3dAutomask 25.2.09, fMRIPrep's build − the repository's build

Bit-identical163,840 of 163,840 voxels equal

the replica against the originalreplica − AFNI 3dAutomask 25.2.09

Bit-identical163,840 of 163,840 voxels equal

the clean room against the originalclean room − AFNI 3dAutomask 25.2.09

Axial slice 20 of 40; the crosshair on the mask's edge. The interactive viewer needs JavaScript; the metrics below are complete without it.

Every comparison, computed in float64 over every voxel of the outputs

Comparison (A − B), verdict and reasonOutputmax |Δ|Values that differRMS of Δmax |Δ| ÷ range1 − NCC
The original against itself: the yardstick
AFNI 3dAutomask 26.3.05 − 25.2.09referenceAFNI 26.3.05 gives AFNI 25.2.09’s bytes on this call: the version step changes nothing here (its two changes, AFNI-03 and AFNI-05, need header slice code 2 or 3dTshift’s -no_detrend).both outputsidentical: 0 of 163,840 values differ
AFNI 3dAutomask 25.2.09, fMRIPrep's build − the repository's buildreferenceTwo builds of AFNI 25.2.09 (fMRIPrep’s image and the repository’s) give the same bytes: AFNI’s own variability on this input is zero.both outputsidentical: 0 of 163,840 values differ
The replica against the original
replica − AFNI 3dAutomask 25.2.09Bit-identicalThe same values, every one.both outputsidentical: 0 of 163,840 values differ
replica − AFNI 3dAutomask 26.3.05Bit-identicalThe same values, every one.both outputsidentical: 0 of 163,840 values differ
replica − AFNI 3dAutomask 25.2.09, fMRIPrep's buildBit-identicalThe same values, every one.both outputsidentical: 0 of 163,840 values differ
The replica against itself: fix and reproduce
replica, fix − reproduceBit-identicalNo correction of the register applies to this input: fix and reproduce give the same bytes.both outputsidentical: 0 of 163,840 values differ
The clean room against the original and the replica
clean room − AFNI 3dAutomask 25.2.09Bit-identicalThe same values, every one.both outputsidentical: 0 of 163,840 values differ
clean room − AFNI 3dAutomask 26.3.05Bit-identicalThe same values, every one.both outputsidentical: 0 of 163,840 values differ
clean room − AFNI 3dAutomask 25.2.09, fMRIPrep's buildBit-identicalThe same values, every one.both outputsidentical: 0 of 163,840 values differ
clean room − replica (both fix)Bit-identicalThe two ports, written independently (one translated from AFNI’s source, one from a behavioural specification), give the same bytes.both outputsidentical: 0 of 163,840 values differ
The clean room against itself: thread counts, fix and reproduce
clean room, fix − reproduceBit-identicalNo correction of the register applies to this input: fix and reproduce give the same bytes.both outputsidentical: 0 of 163,840 values differ
clean room, 6 threads − 1 threadBit-identicalThe clean room’s result does not depend on its thread count: the same bytes at 1 and 6 threads.both outputsidentical: 0 of 163,840 values differ

Δ is A − B: for the masks +1 where only A holds a voxel, −1 where only B does. Computed in float64 over all 163,840 voxels of the outputs, not only those shown. The record’s threshold: every output bit-identical to AFNI’s, with its exit status and messages.

Provenance of this example
  • Data: OpenNeuro ds005040, sub-002, CC0-1.0
  • Inputs: BOLD reference, unifized (sha256 a16170b6e54b3207…); fMRIPrep’s mask (sha256 94be87d836809707…); BOLD (sha256 590cda2ebbff3701…)
  • Runs: figures cache entry afni-automask/fmriprep-ds005040-clip-1, computed 2026-10-23T11:03:14+00:00 with larmorx 71f5223b7836; AFNI_25.2.09, gcc (Ubuntu 11.4.0-1ubuntu1~22.04.3) 11.4.0; AFNI_25.2.09 (Aug 24 2025) [64-bit]; larmorx-gpl 0.0.1; 7f84379b40f399876909ecf7c236105d6b7c43fdec5111fb7b03f8dee204ac10; AFNI_26.3.05; every command, its environment, threads and exit status are in its RUN_INFO.json
  • Runs of the clean room: figures cache entry afni-automask/fmriprep-ds005040-clip-1-original, computed 2026-10-23T21:35:46+00:00 with larmorx 6f0b06b96112 (fix, the default at 1 and 6 threads, reproduce at 1)

Every number above is in afni-automask.json, exported by python -m larmorx_validation evidence afni-automask.

A raw BOLD series

The functional path: with several volumes, 3dAutomask first averages their absolute values, then masks the average.

Worked example

AFNI's default on the raw BOLD series

What this shows: On the raw BOLD series, with no options, the replica and the clean room both give AFNI’s mask, every voxel, in both releases; the clean room on 1 and 6 threads. No correction by default changes this call.

3dAutomask’s usual use: a brain mask of an EPI series. With no options it averages the absolute values of the 63 volumes, then masks the average. The same run as above, before fMRIPrep touched it, given to AFNI 25.2.09, AFNI 26.3.05 (both ports’ reference), fMRIPrep’s AFNI, the replica and the clean room in fix and in reproduce.

Data
OpenNeuro ds005040, sub-002 (CC0-1.0): 64 × 64 × 40 × 63 int16 values × scale factor; shown around the masks, dilated by 2 voxels
Command
3dAutomask -prefix mask.nii.gz sub-002_task-clip_run-1_bold.nii.gz
Programs
AFNI_25.2.09, gcc (Ubuntu 11.4.0-1ubuntu1~22.04.3) 11.4.0; AFNI AFNI_25.2.09 (Aug 24 2025) [64-bit] in fMRIPrep’s image; larmorx-gpl 0.0.1; LARMORX 0.0.1 (the clean room)
Mask
46,005 of 163,840 voxels
Grid
64 × 64 × 40 × 63 voxels of 3 × 3 × 3.3 mm; shown: 53 × 63 × 40 around the brain
AFNI versions
AFNI 26.3.05 (both ports’ reference since 2026-10-29) on the same arguments as AFNI 25.2.09: the same bytes in every output of the call. What changed between the two releases in these programs (header slice code 2, AFNI-03; -no_detrend, AFNI-05; both corrected by LARMORX before AFNI fixed them) does not meet this call.
Inputds005040 sub-002 BOLD, task clip run 1the mean of |value| over its 63 volumes (computed for this page in float64, shown as float32): what 3dAutomask masks
OriginalAFNI 3dAutomask 25.2.09the repository's build
Replicareplicalarmorx, Apache-2.0 with CC BY 4.0's attribution · fix (the default) · against AFNI 3dAutomask 26.3.05, which it follows
Clean roomclean roomlarmorx, Apache-2.0 · fix (the default), 1 thread · against AFNI 3dAutomask 26.3.05, which it follows

Bit-identical163,840 of 163,840 voxels equal

the original against itself: fMRIPrep's build and the repository'sAFNI 3dAutomask 25.2.09, fMRIPrep's build − the repository's build

Bit-identical163,840 of 163,840 voxels equal

the replica against the originalreplica − AFNI 3dAutomask 25.2.09

Bit-identical163,840 of 163,840 voxels equal

the clean room against the originalclean room − AFNI 3dAutomask 25.2.09

Axial slice 19 of 40; the crosshair on the mask's edge. The interactive viewer needs JavaScript; the metrics below are complete without it.

Every comparison, computed in float64 over every voxel of the outputs

Comparison (A − B), verdict and reasonmax |Δ|Values that differRMS of Δmax |Δ| ÷ range1 − NCC
The original against itself: the yardstick
AFNI 3dAutomask 26.3.05 − 25.2.09referenceAFNI 26.3.05 gives AFNI 25.2.09’s bytes on this call: the version step changes nothing here (its two changes, AFNI-03 and AFNI-05, need header slice code 2 or 3dTshift’s -no_detrend).identical: 0 of 163,840 values differ
AFNI 3dAutomask 25.2.09, fMRIPrep's build − the repository's buildreferenceTwo builds of AFNI 25.2.09 (fMRIPrep’s image and the repository’s) give the same bytes: AFNI’s own variability on this input is zero.identical: 0 of 163,840 values differ
The replica against the original
replica − AFNI 3dAutomask 25.2.09Bit-identicalThe same values, every one.identical: 0 of 163,840 values differ
replica − AFNI 3dAutomask 26.3.05Bit-identicalThe same values, every one.identical: 0 of 163,840 values differ
The replica against itself: fix and reproduce
replica, fix − reproduceBit-identicalNo correction of the register applies to this input: fix and reproduce give the same bytes.identical: 0 of 163,840 values differ
The clean room against the original and the replica
clean room − AFNI 3dAutomask 25.2.09Bit-identicalThe same values, every one.identical: 0 of 163,840 values differ
clean room − AFNI 3dAutomask 26.3.05Bit-identicalThe same values, every one.identical: 0 of 163,840 values differ
clean room − replica (both fix)Bit-identicalThe two ports, written independently (one translated from AFNI’s source, one from a behavioural specification), give the same bytes.identical: 0 of 163,840 values differ
The clean room against itself: thread counts, fix and reproduce
clean room, fix − reproduceBit-identicalNo correction of the register applies to this input: fix and reproduce give the same bytes.identical: 0 of 163,840 values differ
clean room, 6 threads − 1 threadBit-identicalThe clean room’s result does not depend on its thread count: the same bytes at 1 and 6 threads.identical: 0 of 163,840 values differ

Δ is A − B: for the masks +1 where only A holds a voxel, −1 where only B does. Computed in float64 over all 163,840 voxels of the outputs, not only those shown. The record’s threshold: every output bit-identical to AFNI’s, with its exit status and messages.

Provenance of this example
  • Data: OpenNeuro ds005040, sub-002, CC0-1.0
  • Inputs: BOLD reference, unifized (sha256 a16170b6e54b3207…); fMRIPrep’s mask (sha256 94be87d836809707…); BOLD (sha256 590cda2ebbff3701…)
  • Runs: figures cache entry afni-automask/fmriprep-ds005040-clip-1, computed 2026-10-23T11:03:14+00:00 with larmorx 71f5223b7836; AFNI_25.2.09, gcc (Ubuntu 11.4.0-1ubuntu1~22.04.3) 11.4.0; AFNI_25.2.09 (Aug 24 2025) [64-bit]; larmorx-gpl 0.0.1; 7f84379b40f399876909ecf7c236105d6b7c43fdec5111fb7b03f8dee204ac10; AFNI_26.3.05; every command, its environment, threads and exit status are in its RUN_INFO.json
  • Runs of the clean room: figures cache entry afni-automask/fmriprep-ds005040-clip-1-original, computed 2026-10-23T21:35:46+00:00 with larmorx 6f0b06b96112 (fix, the default at 1 and 6 threads, reproduce at 1)

Every number above is in afni-automask.json, exported by python -m larmorx_validation evidence afni-automask.

A deliberate correction: -peels 0 (AFNI-43)

3dAutomask’s help says -peels 0 turns the peeling off; AFNI peels once instead (AFNI-43). Both ports’ default, fix, does not peel, and warns; reproduce gives AFNI’s mask.

Worked example

AFNI-43: -peels 0 peels once

What this shows: AFNI peels once at -peels 0, as at -peels 1, in both releases (AFNI-43). In fix, their default, the replica and the clean room do not peel, as AFNI’s help says, and warn; in reproduce both give AFNI’s mask, every voxel.

3dAutomask’s help: “-peels pp = Peel (erode) the mask ‘pp’ times, then unpeel (dilate). Using NN2 neighborhoods, clips off protuberances less than 2*pp voxels thick. Turn off by setting to 0.” AFNI peels once for 0. fMRIPrep’s call with -peels 0 added, given to AFNI 25.2.09, to the replica and to the clean room in fix and in reproduce, beside fMRIPrep’s own call.

Input
the unifized BOLD reference of the first example
Command
3dAutomask -apply_prefix masked.nii.gz -dilate 1 -peels 0 -prefix mask.nii.gz uni_xform.nii.gz
Defect
AFNI-43: -peels 0 peels once; corrected by default, with a warning
Effect
the corrected mask differs from AFNI’s in 343 of 163,840 voxels
AFNI versions
AFNI 26.3.05 (both ports’ reference since 2026-10-29) on the same arguments as AFNI 25.2.09: the same bytes in every output of the call with -peels 0. What changed between the two releases in these programs (header slice code 2, AFNI-03; -no_detrend, AFNI-05; both corrected by LARMORX before AFNI fixed them) does not meet this call.
Inputds005040 sub-002 BOLD reference, unifizedas in the first example
OriginalAFNI 3dAutomask 25.2.09-peels 0 · the repository's build
Clean roomclean roomlarmorx, Apache-2.0 · -peels 0, fix (the default), corrected · against AFNI 3dAutomask 25.2.09
Clean roomclean roomlarmorx, Apache-2.0 · -peels 0, reproduce · against AFNI 3dAutomask 25.2.09

Bit-identical163,840 of 163,840 voxels equal

the original against itself: the defect (-peels 0 and -peels 1)AFNI 3dAutomask 25.2.09: -peels 0 − -peels 1
the clean room against the originalclean room (fix) − AFNI 3dAutomask 25.2.09, -peels 0scale ±1

Bit-identical163,840 of 163,840 voxels equal

the clean room against the originalclean room (reproduce) − AFNI 3dAutomask 25.2.09, -peels 0

Axial slice 35 of 40; the crosshair where the corrected mask differs from AFNI's. The interactive viewer needs JavaScript; the metrics below are complete without it.

Every comparison, computed in float64 over every voxel of the outputs

Comparison (A − B), verdict and reasonOutputmax |Δ|Values that differRMS of Δmax |Δ| ÷ range1 − NCC
The original against itself: the defect
AFNI 3dAutomask 25.2.09: -peels 0 − -peels 1referenceAFNI’s -peels 0 gives its -peels 1 result, every voxel: it peels once whatever 0 says (AFNI-43). AFNI-43both outputsidentical: 0 of 163,840 values differ
The original against itself: the yardstick
AFNI 3dAutomask 26.3.05 − 25.2.09, -peels 0referenceAFNI 26.3.05 gives AFNI 25.2.09’s bytes on this call: the version step changes nothing here (its two changes, AFNI-03 and AFNI-05, need header slice code 2 or 3dTshift’s -no_detrend). AFNI-43 is in both releases.both outputsidentical: 0 of 163,840 values differ
The replica against the original
replica (fix) − AFNI 3dAutomask 25.2.09, -peels 0Deliberate correctionAFNI-43, corrected by default with a warning: 3dAutomask’s help says -peels 0 turns the peeling off, and the default does not peel; reproduce peels once, as AFNI does. AFNI-43Mask1343 of 163,840 (0.209 %)0.045810.00495
Masked input (-apply_prefix)1e39 of 163,840 (0.00549 %)3.920.6696.71e−5
replica (reproduce) − AFNI 3dAutomask 25.2.09, -peels 0Bit-identicalIn reproduce, the replica gives AFNI’s result, defect included. AFNI-43both outputsidentical: 0 of 163,840 values differ
replica (reproduce) − AFNI 3dAutomask 26.3.05, -peels 0Bit-identicalIn reproduce, the replica gives AFNI’s result, defect included. AFNI-43both outputsidentical: 0 of 163,840 values differ
The clean room against the original
clean room (fix) − AFNI 3dAutomask 25.2.09, -peels 0Deliberate correctionAFNI-43, corrected by default with a warning: 3dAutomask’s help says -peels 0 turns the peeling off, and the default does not peel; reproduce peels once, as AFNI does. AFNI-43Mask1343 of 163,840 (0.209 %)0.045810.00495
Masked input (-apply_prefix)1e39 of 163,840 (0.00549 %)3.920.6696.71e−5
clean room (reproduce) − AFNI 3dAutomask 25.2.09, -peels 0Bit-identicalIn reproduce, the clean room gives AFNI’s result, defect included. AFNI-43both outputsidentical: 0 of 163,840 values differ

Δ is A − B: for the masks +1 where only A holds a voxel, −1 where only B does. Computed in float64 over all 163,840 voxels of the outputs, not only those shown. The record’s threshold: every output bit-identical to AFNI’s, with its exit status and messages.

Provenance of this example
  • Data: OpenNeuro ds005040, sub-002, CC0-1.0
  • Inputs: BOLD reference, unifized (sha256 a16170b6e54b3207…); fMRIPrep’s mask (sha256 94be87d836809707…); BOLD (sha256 590cda2ebbff3701…)
  • Runs: figures cache entry afni-automask/fmriprep-ds005040-clip-1, computed 2026-10-23T11:03:14+00:00 with larmorx 71f5223b7836; AFNI_25.2.09, gcc (Ubuntu 11.4.0-1ubuntu1~22.04.3) 11.4.0; AFNI_25.2.09 (Aug 24 2025) [64-bit]; larmorx-gpl 0.0.1; 7f84379b40f399876909ecf7c236105d6b7c43fdec5111fb7b03f8dee204ac10; AFNI_26.3.05; every command, its environment, threads and exit status are in its RUN_INFO.json
  • Runs of the clean room: figures cache entry afni-automask/fmriprep-ds005040-clip-1-original, computed 2026-10-23T21:35:46+00:00 with larmorx 6f0b06b96112 (fix, the default at 1 and 6 threads, reproduce at 1)

Every number above is in afni-automask.json, exported by python -m larmorx_validation evidence afni-automask.

On the data panel: extreme head motion

The first case of the data panel on this page, added beside the examples above: AFNI’s default on a child’s resting-state run with extreme head motion (OpenNeuro ds000256), the whole field of view shown, its noise and the masks’ edges included. Like every panel case that is not one of a program’s key datasets, it is stored as three full-resolution slices through the voxel where the comparison differs most (the field’s centre where nothing differs); its metrics are over every voxel.

Worked example

Data panel: a child's run with extreme head motion

What this shows: In fix, the default, the replica and the clean room both give AFNI’s bytes (the clean room on 1 and 3 threads). No correction by default changes this call, in either port.

The first case of the data panel on this page: AFNI’s default (the mean of the absolute values over the 164 volumes, then its mask) on a 7.5-year-old’s resting-state run whose head moved 4.97 mm per volume on average (the dataset’s MRIQC), given to AFNI 25.2.09, fMRIPrep’s AFNI, the replica and the clean room in fix and in reproduce. The whole field of view is shown, its noise and the mask’s edges included; as a panel case beside the examples above, it is stored as three full-resolution slices.

Data
OpenNeuro ds000256 (Greene et al., 2018), snapshot 00002, sub-CTS203, task restfixationnofeedback (PDDL-1.0): 64 × 64 × 32 × 164 int16 values; the whole field of view shown
Command
3dAutomask -prefix mask.nii.gz sub-CTS203_task-restfixationnofeedback_bold.nii.gz
Programs
AFNI_25.2.09, gcc (Ubuntu 11.4.0-1ubuntu1~22.04.3) 11.4.0; AFNI AFNI_25.2.09 (Aug 24 2025) [64-bit] in fMRIPrep’s image; larmorx-gpl 0.0.1; LARMORX 0.0.1 (the clean room)
Mask
23,168 of 131,072 voxels
Grid
64 × 64 × 32 × 164 voxels of 4 × 4 × 4 mm; shown: the whole field of view (64 × 64 × 32), only margins where every array is 0 left out
AFNI versions
AFNI 26.3.05 (both ports’ reference since 2026-10-29) on the same arguments as AFNI 25.2.09: the same bytes in every output of the call. What changed between the two releases in these programs (header slice code 2, AFNI-03; -no_detrend, AFNI-05; both corrected by LARMORX before AFNI fixed them) does not meet this call.
Inputds000256 sub-CTS203 BOLD, rest (no feedback)the mean of |value| over its 164 volumes (computed for this page in float64, shown as float32): what 3dAutomask masks
OriginalAFNI 3dAutomask 25.2.09the repository's build
Replicareplicalarmorx, Apache-2.0 with CC BY 4.0's attribution · fix (the default) · against AFNI 3dAutomask 26.3.05, which it follows
Clean roomclean roomlarmorx, Apache-2.0 · fix (the default), 1 thread · against AFNI 3dAutomask 26.3.05, which it follows

Bit-identical131,072 of 131,072 voxels equal

the original against itself: fMRIPrep's build and the repository'sAFNI 3dAutomask 25.2.09, fMRIPrep's build − the repository's build

Bit-identical131,072 of 131,072 voxels equal

the replica against the originalreplica − AFNI 3dAutomask 25.2.09

Bit-identical131,072 of 131,072 voxels equal

the clean room against the originalclean room − AFNI 3dAutomask 25.2.09

Axial slice 16 of 32; the crosshair at the centre of the field of view; the three slices through it are stored at full resolution. The interactive viewer needs JavaScript; the metrics below are complete without it.

Stored for this example: three full-resolution slices, one per plane, through display voxel (32, 32, 16): where the main comparison differs most, or the field's centre where nothing differs; every metric is computed over every voxel of the files.

Every comparison, computed in float64 over every voxel of the outputs

Comparison (A − B), verdict and reasonmax |Δ|Values that differRMS of Δmax |Δ| ÷ range1 − NCC
The original against itself: the yardstick
AFNI 3dAutomask 26.3.05 − 25.2.09referenceAFNI 26.3.05 gives AFNI 25.2.09’s bytes on this call: the version step changes nothing here (its two changes, AFNI-03 and AFNI-05, need header slice code 2 or 3dTshift’s -no_detrend).identical: 0 of 131,072 values differ
AFNI 3dAutomask 25.2.09, fMRIPrep's build − the repository's buildreferenceTwo builds of AFNI 25.2.09 (fMRIPrep’s image and the repository’s) give the same bytes: AFNI’s own variability on this input is zero.identical: 0 of 131,072 values differ
The replica against the original
replica − AFNI 3dAutomask 25.2.09Bit-identicalThe same values, every one.identical: 0 of 131,072 values differ
replica − AFNI 3dAutomask 26.3.05Bit-identicalThe same values, every one.identical: 0 of 131,072 values differ
The replica against itself: fix and reproduce
replica, fix − reproduceBit-identicalNo correction of the register applies to this input: fix and reproduce give the same bytes.identical: 0 of 131,072 values differ
The clean room against the original and the replica
clean room − AFNI 3dAutomask 25.2.09Bit-identicalThe same values, every one.identical: 0 of 131,072 values differ
clean room − AFNI 3dAutomask 26.3.05Bit-identicalThe same values, every one.identical: 0 of 131,072 values differ
clean room − replica (both fix)Bit-identicalThe two ports, written independently (one translated from AFNI’s source, one from a behavioural specification), give the same bytes.identical: 0 of 131,072 values differ
The clean room against itself: thread counts, fix and reproduce
clean room, fix − reproduceBit-identicalNo correction of the register applies to this input: fix and reproduce give the same bytes.identical: 0 of 131,072 values differ
clean room, 3 threads − 1 threadBit-identicalThe clean room’s result does not depend on its thread count: the same bytes at 1 and 3 threads.identical: 0 of 131,072 values differ

Δ is A − B: for the masks +1 where only A holds a voxel, −1 where only B does. Computed in float64 over all 131,072 voxels of the outputs, not only those shown. The record’s threshold: every output bit-identical to AFNI’s, with its exit status and messages.

Provenance of this example
  • Data: OpenNeuro ds000256 (Greene et al., 2018), snapshot 00002, sub-CTS203, task restfixationnofeedback, PDDL-1.0
  • Inputs: BOLD (sha256 de3e3a8c3e2d4f6b…)
  • Runs: figures cache entry afni-automask/panel-ds000256-CTS203-nofeedback, computed 2026-10-24T15:31:29+00:00 with larmorx 569ea25214b5; AFNI_25.2.09, gcc (Ubuntu 11.4.0-1ubuntu1~22.04.3) 11.4.0; AFNI_25.2.09 (Aug 24 2025) [64-bit]; larmorx-gpl 0.0.1; 7f84379b40f399876909ecf7c236105d6b7c43fdec5111fb7b03f8dee204ac10; “larmorx 0.0.1 (the clean-room original)”, as the cache entry records it; AFNI_26.3.05; every command, its environment, threads and exit status are in its RUN_INFO.json

Every number above is in afni-automask.json, exported by python -m larmorx_validation evidence afni-automask.

Every case

Each case of the two records, with what it tests, its command line, each port’s verdict and the reason. Search, filter by category or verdict (or select a segment of a chart above), sort by a column; #case=<name> links to one case.

131 cases

CaseCategoryWhat it testsclean roomreplicaChecks
options/defaultoptionsthe defaults (gradual clip level 0.5, 1 peel, NN2)
command line3dAutomask -prefix mask.nii <generated: head-f32>
Bit-identicalBit-identical8/8
options/fmriprep-calloptionsfMRIPrep's options: -apply_prefix, -dilate 1
command line3dAutomask -apply_prefix masked.nii -dilate 1 -prefix mask.nii <generated: head-f32>
Bit-identicalBit-identical10/10
options/clfrac-0.1options-clfrac 0.1
command line3dAutomask -clfrac 0.1 -prefix mask.nii <generated: head-f32>
Bit-identicalBit-identical8/8
options/clfrac-0.3options-clfrac 0.3
command line3dAutomask -clfrac 0.3 -prefix mask.nii <generated: head-f32>
Bit-identicalBit-identical8/8
options/clfrac-0.9options-clfrac 0.9
command line3dAutomask -clfrac 0.9 -prefix mask.nii <generated: head-f32>
Bit-identicalBit-identical8/8
options/mfracoptions-mfrac, the same as -clfrac
command line3dAutomask -mfrac 0.35 -prefix mask.nii <generated: head-f32>
Bit-identicalBit-identical7/7
options/nogradoptions-nograd: one clip level
command line3dAutomask -nograd -prefix mask.nii <generated: head-f32>
Bit-identicalBit-identical8/8
options/nogr-prefixoptions-nogr: AFNI compares 5 characters
command line3dAutomask -nogrxyz -prefix mask.nii <generated: head-f32>
Bit-identicalBit-identical7/7
options/peels-2options-peels 2
command line3dAutomask -peels 2 -prefix mask.nii <generated: head-f32>
Bit-identicalBit-identical8/8
options/peels-3options-peels 3
command line3dAutomask -peels 3 -prefix mask.nii <generated: head-f32>
Bit-identicalBit-identical8/8
options/nbhrs-6options-nbhrs 6 (AFNI uses 17 outside 6..26, AFNI-41)
command line3dAutomask -nbhrs 6 -prefix mask.nii <generated: head-f32>
Bit-identicalBit-identical7/7
options/nbhrs-15options-nbhrs 15 (AFNI uses 17 outside 6..26, AFNI-41)
command line3dAutomask -nbhrs 15 -prefix mask.nii <generated: head-f32>
Bit-identicalBit-identical7/7
options/nbhrs-18options-nbhrs 18 (AFNI uses 17 outside 6..26, AFNI-41)
command line3dAutomask -nbhrs 18 -prefix mask.nii <generated: head-f32>
Bit-identicalBit-identical7/7
options/nbhrs-26options-nbhrs 26 (AFNI uses 17 outside 6..26, AFNI-41)
command line3dAutomask -nbhrs 26 -prefix mask.nii <generated: head-f32>
Bit-identicalBit-identical7/7
options/nbhrs-40options-nbhrs 40 (AFNI uses 17 outside 6..26, AFNI-41)
command line3dAutomask -nbhrs 40 -prefix mask.nii <generated: head-f32>
Deliberate correction

Refused by default (AFNI-41, decided 2026-10-23, rule B: the input makes no sense and AFNI silently does something arbitrary): the help gives -nbhrs a range of 6 to 26, but AFNI accepts any value and peels as with 17 (or as -NN1/2/3 say). The clean room’s default stops with a fatal error naming the entry (exit status 1); reproduce gives AFNI’s output, which the record compares with AFNI’s.

Deliberate correction

Refused by default (AFNI-41, decided 2026-10-23, rule B: the input makes no sense and AFNI silently does something arbitrary): the help gives -nbhrs a range of 6 to 26, but AFNI accepts any value and peels as with 17 (or as -NN1/2/3 say). The replica’s default stops with a fatal error naming the entry (exit status 1); reproduce gives AFNI’s output, which the record compares with AFNI’s.

7/7
options/nbhrs-5options-nbhrs 5: outside 6..26, refused by default (AFNI-41)
command line3dAutomask -nbhrs 5 -prefix mask.nii <generated: head-f32>
Deliberate correction

Refused by default (AFNI-41, decided 2026-10-23, rule B: the input makes no sense and AFNI silently does something arbitrary): the help gives -nbhrs a range of 6 to 26, but AFNI accepts any value and peels as with 17 (or as -NN1/2/3 say). The clean room’s default stops with a fatal error naming the entry (exit status 1); reproduce gives AFNI’s output, which the record compares with AFNI’s.

Deliberate correction

Refused by default (AFNI-41, decided 2026-10-23, rule B: the input makes no sense and AFNI silently does something arbitrary): the help gives -nbhrs a range of 6 to 26, but AFNI accepts any value and peels as with 17 (or as -NN1/2/3 say). The replica’s default stops with a fatal error naming the entry (exit status 1); reproduce gives AFNI’s output, which the record compares with AFNI’s.

7/7
options/nbhrs-27options-nbhrs 27: outside 6..26, refused by default (AFNI-41)
command line3dAutomask -nbhrs 27 -prefix mask.nii <generated: head-f32>
Deliberate correction

Refused by default (AFNI-41, decided 2026-10-23, rule B: the input makes no sense and AFNI silently does something arbitrary): the help gives -nbhrs a range of 6 to 26, but AFNI accepts any value and peels as with 17 (or as -NN1/2/3 say). The clean room’s default stops with a fatal error naming the entry (exit status 1); reproduce gives AFNI’s output, which the record compares with AFNI’s.

Deliberate correction

Refused by default (AFNI-41, decided 2026-10-23, rule B: the input makes no sense and AFNI silently does something arbitrary): the help gives -nbhrs a range of 6 to 26, but AFNI accepts any value and peels as with 17 (or as -NN1/2/3 say). The replica’s default stops with a fatal error naming the entry (exit status 1); reproduce gives AFNI’s output, which the record compares with AFNI’s.

7/7
options/nbhrs-0options-nbhrs 0: outside 6..26, refused by default (AFNI-41)
command line3dAutomask -nbhrs 0 -prefix mask.nii <generated: head-f32>
Deliberate correction

Refused by default (AFNI-41, decided 2026-10-23, rule B: the input makes no sense and AFNI silently does something arbitrary): the help gives -nbhrs a range of 6 to 26, but AFNI accepts any value and peels as with 17 (or as -NN1/2/3 say). The clean room’s default stops with a fatal error naming the entry (exit status 1); reproduce gives AFNI’s output, which the record compares with AFNI’s.

Deliberate correction

Refused by default (AFNI-41, decided 2026-10-23, rule B: the input makes no sense and AFNI silently does something arbitrary): the help gives -nbhrs a range of 6 to 26, but AFNI accepts any value and peels as with 17 (or as -NN1/2/3 say). The replica’s default stops with a fatal error naming the entry (exit status 1); reproduce gives AFNI’s output, which the record compares with AFNI’s.

7/7
options/nbhrs-40-then-NN1options-nbhrs 40 then -NN1: AFNI peels as -NN1 says; refused by default (AFNI-41)
command line3dAutomask -nbhrs 40 -NN1 -prefix mask.nii <generated: head-f32>
Deliberate correction

Refused by default (AFNI-41, decided 2026-10-23, rule B: the input makes no sense and AFNI silently does something arbitrary): the help gives -nbhrs a range of 6 to 26, but AFNI accepts any value and peels as with 17 (or as -NN1/2/3 say). The clean room’s default stops with a fatal error naming the entry (exit status 1); reproduce gives AFNI’s output, which the record compares with AFNI’s.

Deliberate correction

Refused by default (AFNI-41, decided 2026-10-23, rule B: the input makes no sense and AFNI silently does something arbitrary): the help gives -nbhrs a range of 6 to 26, but AFNI accepts any value and peels as with 17 (or as -NN1/2/3 say). The replica’s default stops with a fatal error naming the entry (exit status 1); reproduce gives AFNI’s output, which the record compares with AFNI’s.

7/7
options/NN1-dilate-erode-depthoptions-NN1 with -dilate 2, -erode 1 and -depth (and the peeling count it sets, AFNI-42)
command line3dAutomask -NN1 -dilate 2 -erode 1 -depth depth.nii -prefix mask.nii <generated: head-f32>
Bit-identicalBit-identical9/9
options/NN2-dilate-erode-depthoptions-NN2 with -dilate 2, -erode 1 and -depth (and the peeling count it sets, AFNI-42)
command line3dAutomask -NN2 -dilate 2 -erode 1 -depth depth.nii -prefix mask.nii <generated: head-f32>
Bit-identicalBit-identical9/9
options/NN3-dilate-erode-depthoptions-NN3 with -dilate 2, -erode 1 and -depth (and the peeling count it sets, AFNI-42)
command line3dAutomask -NN3 -dilate 2 -erode 1 -depth depth.nii -prefix mask.nii <generated: head-f32>
Bit-identicalBit-identical9/9
options/nbhrs-then-NN2options-nbhrs 12 then -NN2: -NN2 sets 18 (AFNI-42)
command line3dAutomask -nbhrs 12 -NN2 -prefix mask.nii <generated: head-f32>
Bit-identicalBit-identical7/7
options/eclipoptions-eclip
command line3dAutomask -eclip -prefix mask.nii <generated: head-f32>
Bit-identicalBit-identical8/8
options/eclip-nograd-clfracoptions-eclip -nograd -clfrac 0.2
command line3dAutomask -eclip -nograd -clfrac 0.2 -prefix mask.nii <generated: head-f32>
Bit-identicalBit-identical7/7
options/dilate-3options-dilate 3
command line3dAutomask -dilate 3 -prefix mask.nii <generated: head-f32>
Bit-identicalBit-identical7/7
options/erode-2options-erode 2
command line3dAutomask -erode 2 -prefix mask.nii <generated: head-f32>
Bit-identicalBit-identical8/8
options/depthoptions-depth alone
command line3dAutomask -depth depth.nii -prefix mask.nii <generated: head-f32>
Bit-identicalBit-identical10/10
options/apply-onlyoptions-apply_prefix without -prefix: the only output
command line3dAutomask -apply_prefix masked.nii <generated: head-f32>
Bit-identicalBit-identical7/7
options/quietoptions-q
command line3dAutomask -q -prefix mask.nii <generated: head-f32>
Bit-identicalBit-identical7/7
options/SI-head-f32options-SI 60 along z, superior last
command line3dAutomask -SI 60 -prefix mask.nii <generated: head-f32>
Bit-identicalBit-identical8/8
options/SI-head-superior-firstoptions-SI 60 along z, superior first
command line3dAutomask -SI 60 -prefix mask.nii <generated: head-superior-first>
Bit-identicalBit-identical7/7
options/SI-head-coronaloptions-SI 60 along y
command line3dAutomask -SI 60 -prefix mask.nii <generated: head-coronal>
Bit-identicalBit-identical7/7
options/SI-head-lasoptions-SI 60 along z, LAS storage
command line3dAutomask -SI 60 -prefix mask.nii <generated: head-las>
Bit-identicalBit-identical7/7
options/SI-everythingoptions-SI 1: clipping beyond the mask
command line3dAutomask -SI 1 -prefix mask.nii <generated: head-f32>
Bit-identicalBit-identical7/7
options/t2-contrastoptionsthe T2-like phantom
command line3dAutomask -prefix mask.nii <generated: head-t2-f32>
Bit-identicalBit-identical7/7
options/bigoptionsa volume of 1.2 million voxels
command line3dAutomask -apply_prefix masked.nii.gz -prefix mask.nii.gz <generated: big>
Bit-identicalBit-identical9/9
datatypes/head-i16data typeshead-i16: the mask, the masked input and the depth
command line3dAutomask -apply_prefix masked.nii -depth depth.nii -prefix mask.nii <generated: head-i16>
Bit-identicalBit-identical11/11
datatypes/head-u8data typeshead-u8: the mask, the masked input and the depth
command line3dAutomask -apply_prefix masked.nii -depth depth.nii -prefix mask.nii <generated: head-u8>
Bit-identicalBit-identical11/11
datatypes/head-i16-slope2data typeshead-i16-slope2: the mask, the masked input and the depth
command line3dAutomask -apply_prefix masked.nii -depth depth.nii -prefix mask.nii <generated: head-i16-slope2>
Bit-identicalBit-identical11/11
datatypes/head-i16-slope1data typeshead-i16-slope1: the mask, the masked input and the depth
command line3dAutomask -apply_prefix masked.nii -depth depth.nii -prefix mask.nii <generated: head-i16-slope1>
Bit-identicalBit-identical11/11
datatypes/head-f64data typeshead-f64: the mask, the masked input and the depth
command line3dAutomask -apply_prefix masked.nii -depth depth.nii -prefix mask.nii <generated: head-f64>
Bit-identicalBit-identical11/11
datatypes/head-i8data typeshead-i8: the mask, the masked input and the depth
command line3dAutomask -apply_prefix masked.nii -depth depth.nii -prefix mask.nii <generated: head-i8>
Bit-identicalBit-identical11/11
datatypes/head-u16data typeshead-u16: the mask, the masked input and the depth
command line3dAutomask -apply_prefix masked.nii -depth depth.nii -prefix mask.nii <generated: head-u16>
Bit-identicalBit-identical11/11
datatypes/head-i32data typeshead-i32: the mask, the masked input and the depth
command line3dAutomask -apply_prefix masked.nii -depth depth.nii -prefix mask.nii <generated: head-i32>
Bit-identicalBit-identical11/11
datatypes/head-nandata typeshead-nan: the mask, the masked input and the depth
command line3dAutomask -apply_prefix masked.nii -depth depth.nii -prefix mask.nii <generated: head-nan>
Bit-identicalBit-identical11/11
datatypes/head-obliquedata typeshead-oblique: the mask, the masked input and the depth
command line3dAutomask -apply_prefix masked.nii -depth depth.nii -prefix mask.nii <generated: head-oblique>
Bit-identicalBit-identical11/11
datatypes/head-qform-onlydata typeshead-qform-only: the mask, the masked input and the depth
command line3dAutomask -apply_prefix masked.nii -depth depth.nii -prefix mask.nii <generated: head-qform-only>
Bit-identicalBit-identical11/11
datatypes/head-sform-onlydata typeshead-sform-only: the mask, the masked input and the depth
command line3dAutomask -apply_prefix masked.nii -depth depth.nii -prefix mask.nii <generated: head-sform-only>
Bit-identicalBit-identical11/11
datatypes/head-lasdata typeshead-las: the mask, the masked input and the depth
command line3dAutomask -apply_prefix masked.nii -depth depth.nii -prefix mask.nii <generated: head-las>
Bit-identicalBit-identical11/11
defects/head-f32-slope-interdefects of AFNIfloat32 with scl_slope and scl_inter (AFNI-02)
command line3dAutomask -apply_prefix masked.nii -prefix mask.nii <generated: head-f32-slope-inter>
Bit-identical

in fix, its default, the clean room corrects AFNI-02 and names it: the header or the messages change (the record’s list of what fix changed)

Bit-identical

in fix, its default, the replica corrects AFNI-02 and names it: the header or the messages change (the record’s list of what fix changed)

9/9
defects/head-metredefects of AFNIa grid in metres (AFNI-07)
command line3dAutomask -prefix mask.nii <generated: head-metre>
Bit-identical

in fix, its default, the clean room corrects AFNI-07 and names it: the header or the messages change (the record’s list of what fix changed)

Bit-identical

in fix, its default, the replica corrects AFNI-07 and names it: the header or the messages change (the record’s list of what fix changed)

7/7
defects/head-nearly-plumbdefects of AFNIa grid 0.009 degrees from plumb (AFNI-16)
command line3dAutomask -prefix mask.nii <generated: head-nearly-plumb>
Bit-identical

in fix, its default, the clean room corrects AFNI-16 and names it: the header or the messages change (the record’s list of what fix changed)

Bit-identical

in fix, its default, the replica corrects AFNI-16 and names it: the header or the messages change (the record’s list of what fix changed)

7/7
defects/nifti-scale-nodefects of AFNIAFNI_NIFTI_SCALE=NO on int16 with an intercept (AFNI-23)
command lineAFNI_NIFTI_SCALE=NO 3dAutomask -apply_prefix masked.nii -prefix mask.nii <generated: head-i16-intercept>
Bit-identical

in fix, its default, the clean room corrects AFNI-23 and names it: 163,840 values change (the record’s list of what fix changed)

Bit-identical

in fix, its default, the replica corrects AFNI-23 and names it: 163,840 values change (the record’s list of what fix changed)

9/9
defects/peels-0defects of AFNI-peels 0 (AFNI-43: AFNI peels once)
command line3dAutomask -peels 0 -prefix mask.nii <generated: head-f32>
Bit-identical

in fix, its default, the clean room corrects AFNI-43 and names it: the header or the messages change (the record’s list of what fix changed)

Bit-identical

in fix, its default, the replica corrects AFNI-43 and names it: the header or the messages change (the record’s list of what fix changed)

8/8
defects/peels-0-fmriprepdefects of AFNI-peels 0 added to fMRIPrep's call on its BOLD reference of ds005040 (AFNI-43)
command line3dAutomask -apply_prefix masked.nii.gz -dilate 1 -peels 0 -prefix mask.nii.gz <fMRIPrep ds005040-task_clip_run_1: unifized BOLD reference>
Bit-identical

in fix, its default, the clean room corrects AFNI-43 and names it: 352 values change (the record’s list of what fix changed)

Bit-identical

in fix, its default, the replica corrects AFNI-43 and names it: 352 values change (the record’s list of what fix changed)

9/9
defects/negative-slope-4ddefects of AFNIa 4D series with scl_slope -2 (AFNI-01)
command line3dAutomask -apply_prefix masked.nii -prefix mask.nii <generated: series-negative>
Bit-identical

in fix, its default, the clean room corrects AFNI-01 and names it: the header or the messages change (the record’s list of what fix changed)

Bit-identical

in fix, its default, the replica corrects AFNI-01 and names it: the header or the messages change (the record’s list of what fix changed)

9/9
dimensions/series-i16dimensionsa 4D int16 series: the mean of |values|
command line3dAutomask -apply_prefix masked.nii -prefix mask.nii <generated: series-i16>
Bit-identicalBit-identical9/9
dimensions/series-f32dimensionsa 4D float32 series
command line3dAutomask -apply_prefix masked.nii -prefix mask.nii <generated: series-f32>
Bit-identicalBit-identical9/9
dimensions/series-slopedimensionsa 4D series with scl_slope 2
command line3dAutomask -apply_prefix masked.nii -prefix mask.nii <generated: series-slope>
Bit-identicalBit-identical9/9
dimensions/series-onedimensionsa 4D file of one volume
command line3dAutomask -apply_prefix masked.nii -prefix mask.nii <generated: series-one>
Bit-identicalBit-identical9/9
dimensions/slicedimensionsa 2D slice (64 x 64 x 1): AFNI's 2D path
command line3dAutomask -apply_prefix masked.nii -prefix mask.nii <generated: slice>
Bit-identicalBit-identical9/9
dimensions/two-slicesdimensionstwo slices (64 x 64 x 2)
command line3dAutomask -apply_prefix masked.nii -prefix mask.nii <generated: two-slices>
Bit-identicalBit-identical9/9
dimensions/tinydimensions2 x 2 x 2 voxels: nothing to cluster
command line3dAutomask -apply_prefix masked.nii -prefix mask.nii <generated: tiny>
Bit-identicalBit-identical9/9
dimensions/zerosdimensionsan image of zeros: the clip level is 0
command line3dAutomask -apply_prefix masked.nii -prefix mask.nii <generated: zeros>
Bit-identicalBit-identical9/9
dimensions/constantdimensionsa constant image
command line3dAutomask -apply_prefix masked.nii -prefix mask.nii <generated: constant>
Bit-identicalBit-identical9/9
errors/slice-smallerrorsa 2D slice smaller than 16 x 16: no mask
command line3dAutomask -prefix mask.nii <generated: slice-small>
Rejected by both

AFNI exit 1: ** FATAL ERROR: Mask creation fails for unknown reasons!; the clean room exit 1: ** FATAL ERROR: Mask creation fails for unknown reasons! (the same messages)

Rejected by both

AFNI exit 1: ** FATAL ERROR: Mask creation fails for unknown reasons!; the replica exit 1: ** FATAL ERROR: Mask creation fails for unknown reasons! (the same messages)

4/4
errors/rowerrorsone row (64 x 1 x 1): no mask
command line3dAutomask -prefix mask.nii <generated: row>
Rejected by both

AFNI exit 1: ** FATAL ERROR: Mask creation fails for unknown reasons!; the clean room exit 1: ** FATAL ERROR: Mask creation fails for unknown reasons! (the same messages)

Rejected by both

AFNI exit 1: ** FATAL ERROR: Mask creation fails for unknown reasons!; the replica exit 1: ** FATAL ERROR: Mask creation fails for unknown reasons! (the same messages)

4/4
errors/missing-value-peelserrors-peels as the last argument: AFNI crashes (AFNI-40)
command line3dAutomask -q -peels
Expected divergence

AFNI crashes (exit 1, Fatal Signal 11 (SIGSEGV) received); the clean room refuses (exit 1: ** FATAL ERROR: -peels needs a value: AFNI crashes (Fatal Signal 11) when an option’s value is missing (AFNI-40)), after the same messages

Expected divergence

AFNI crashes (exit 1, Fatal Signal 11 (SIGSEGV) received); the replica refuses (exit 1: ** FATAL ERROR: -peels needs a value: AFNI crashes (Fatal Signal 11) when an option’s value is missing (AFNI-40)), after the same messages

3/3
errors/missing-value-nbhrserrors-nbhrs as the last argument: AFNI crashes (AFNI-40)
command line3dAutomask -q -nbhrs
Expected divergence

AFNI crashes (exit 1, Fatal Signal 11 (SIGSEGV) received); the clean room refuses (exit 1: ** FATAL ERROR: -nbhrs needs a value: AFNI crashes (Fatal Signal 11) when an option’s value is missing (AFNI-40)), after the same messages

Expected divergence

AFNI crashes (exit 1, Fatal Signal 11 (SIGSEGV) received); the replica refuses (exit 1: ** FATAL ERROR: -nbhrs needs a value: AFNI crashes (Fatal Signal 11) when an option’s value is missing (AFNI-40)), after the same messages

3/3
errors/missing-value-clfracerrors-clfrac as the last argument: AFNI crashes (AFNI-40)
command line3dAutomask -q -clfrac
Expected divergence

AFNI crashes (exit 1, Fatal Signal 11 (SIGSEGV) received); the clean room refuses (exit 1: ** FATAL ERROR: -clfrac needs a value: AFNI crashes (Fatal Signal 11) when an option’s value is missing (AFNI-40)), after the same messages

Expected divergence

AFNI crashes (exit 1, Fatal Signal 11 (SIGSEGV) received); the replica refuses (exit 1: ** FATAL ERROR: -clfrac needs a value: AFNI crashes (Fatal Signal 11) when an option’s value is missing (AFNI-40)), after the same messages

3/3
errors/missing-value-SIerrors-SI as the last argument: AFNI crashes (AFNI-40)
command line3dAutomask -q -SI
Expected divergence

AFNI crashes (exit 1, Fatal Signal 11 (SIGSEGV) received); the clean room refuses (exit 1: ** FATAL ERROR: -SI needs a value: AFNI crashes (Fatal Signal 11) when an option’s value is missing (AFNI-40)), after the same messages

Expected divergence

AFNI crashes (exit 1, Fatal Signal 11 (SIGSEGV) received); the replica refuses (exit 1: ** FATAL ERROR: -SI needs a value: AFNI crashes (Fatal Signal 11) when an option’s value is missing (AFNI-40)), after the same messages

3/3
errors/missing-value-dilateerrors-dilate as the last argument: AFNI crashes (AFNI-40)
command line3dAutomask -q -dilate
Expected divergence

AFNI crashes (exit 1, Fatal Signal 11 (SIGSEGV) received); the clean room refuses (exit 1: ** FATAL ERROR: -dilate needs a value: AFNI crashes (Fatal Signal 11) when an option’s value is missing (AFNI-40)), after the same messages

Expected divergence

AFNI crashes (exit 1, Fatal Signal 11 (SIGSEGV) received); the replica refuses (exit 1: ** FATAL ERROR: -dilate needs a value: AFNI crashes (Fatal Signal 11) when an option’s value is missing (AFNI-40)), after the same messages

3/3
errors/missing-value-erodeerrors-erode as the last argument: AFNI crashes (AFNI-40)
command line3dAutomask -q -erode
Expected divergence

AFNI crashes (exit 1, Fatal Signal 11 (SIGSEGV) received); the clean room refuses (exit 1: ** FATAL ERROR: -erode needs a value: AFNI crashes (Fatal Signal 11) when an option’s value is missing (AFNI-40)), after the same messages

Expected divergence

AFNI crashes (exit 1, Fatal Signal 11 (SIGSEGV) received); the replica refuses (exit 1: ** FATAL ERROR: -erode needs a value: AFNI crashes (Fatal Signal 11) when an option’s value is missing (AFNI-40)), after the same messages

3/3
errors/missing-name-prefixerrors-prefix as the last argument: '(null) is illegal'
command line3dAutomask -q -prefix
Rejected by both

AFNI exit 1: ** FATAL ERROR: -prefix (null) is illegal!; the clean room exit 1: ** FATAL ERROR: -prefix (null) is illegal! (the same messages)

Rejected by both

AFNI exit 1: ** FATAL ERROR: -prefix (null) is illegal!; the replica exit 1: ** FATAL ERROR: -prefix (null) is illegal! (the same messages)

4/4
errors/missing-name-deptherrors-depth as the last argument: '(null) is illegal'
command line3dAutomask -q -depth
Rejected by both

AFNI exit 1: ** FATAL ERROR: -depth (null) is illegal!; the clean room exit 1: ** FATAL ERROR: -depth (null) is illegal! (the same messages)

Rejected by both

AFNI exit 1: ** FATAL ERROR: -depth (null) is illegal!; the replica exit 1: ** FATAL ERROR: -depth (null) is illegal! (the same messages)

4/4
errors/missing-name-apply_prefixerrors-apply_prefix as the last argument: '(null) is illegal'
command line3dAutomask -q -apply_prefix
Rejected by both

AFNI exit 1: ** FATAL ERROR: -apply_prefix (null) is illegal!; the clean room exit 1: ** FATAL ERROR: -apply_prefix (null) is illegal! (the same messages)

Rejected by both

AFNI exit 1: ** FATAL ERROR: -apply_prefix (null) is illegal!; the replica exit 1: ** FATAL ERROR: -apply_prefix (null) is illegal! (the same messages)

4/4
errors/clfrac-higherrors-clfrac 0.95
command line3dAutomask -clfrac 0.95 <generated: head-f32>
Rejected by both

AFNI exit 1: ** FATAL ERROR: -clfrac value 0.950000 is illegal!; the clean room exit 1: ** FATAL ERROR: -clfrac value 0.950000 is illegal! (the same messages)

Rejected by both

AFNI exit 1: ** FATAL ERROR: -clfrac value 0.950000 is illegal!; the replica exit 1: ** FATAL ERROR: -clfrac value 0.950000 is illegal! (the same messages)

4/4
errors/clfrac-lowerrors-clfrac 0.05
command line3dAutomask -clfrac 0.05 <generated: head-f32>
Rejected by both

AFNI exit 1: ** FATAL ERROR: -clfrac value 0.050000 is illegal!; the clean room exit 1: ** FATAL ERROR: -clfrac value 0.050000 is illegal! (the same messages)

Rejected by both

AFNI exit 1: ** FATAL ERROR: -clfrac value 0.050000 is illegal!; the replica exit 1: ** FATAL ERROR: -clfrac value 0.050000 is illegal! (the same messages)

4/4
errors/SI-negativeerrors-SI -3
command line3dAutomask -SI -3 <generated: head-f32>
Rejected by both

AFNI exit 1: ** FATAL ERROR: -SI value -3.000000 is illegal!; the clean room exit 1: ** FATAL ERROR: -SI value -3.000000 is illegal! (the same messages)

Rejected by both

AFNI exit 1: ** FATAL ERROR: -SI value -3.000000 is illegal!; the replica exit 1: ** FATAL ERROR: -SI value -3.000000 is illegal! (the same messages)

4/4
errors/dilate-negativeerrors-dilate -1
command line3dAutomask -dilate -1 <generated: head-f32>
Rejected by both

AFNI exit 1: ** FATAL ERROR: -dilate -1 is illegal!; the clean room exit 1: ** FATAL ERROR: -dilate -1 is illegal! (the same messages)

Rejected by both

AFNI exit 1: ** FATAL ERROR: -dilate -1 is illegal!; the replica exit 1: ** FATAL ERROR: -dilate -1 is illegal! (the same messages)

4/4
errors/erode-negativeerrors-erode -2
command line3dAutomask -erode -2 <generated: head-f32>
Rejected by both

AFNI exit 1: ** FATAL ERROR: -erode -2 is illegal!; the clean room exit 1: ** FATAL ERROR: -erode -2 is illegal! (the same messages)

Rejected by both

AFNI exit 1: ** FATAL ERROR: -erode -2 is illegal!; the replica exit 1: ** FATAL ERROR: -erode -2 is illegal! (the same messages)

4/4
errors/unknown-optionerrorsan unknown option
command line3dAutomask -bogus <generated: head-f32>
Rejected by both

AFNI exit 1: ** FATAL ERROR: ILLEGAL option: -bogus; the clean room exit 1: ** FATAL ERROR: ILLEGAL option: -bogus (the same messages)

Rejected by both

AFNI exit 1: ** FATAL ERROR: ILLEGAL option: -bogus; the replica exit 1: ** FATAL ERROR: ILLEGAL option: -bogus (the same messages)

4/4
errors/no-dataseterrorsno dataset
command line3dAutomask -prefix mask.nii
Rejected by both

AFNI exit 1: ** FATAL ERROR: Can’t open dataset ‘(null)’; the clean room exit 1: ** FATAL ERROR: Can’t open dataset ‘(null)’ (the same messages)

Rejected by both

AFNI exit 1: ** FATAL ERROR: Can’t open dataset ‘(null)’; the replica exit 1: ** FATAL ERROR: Can’t open dataset ‘(null)’ (the same messages)

4/4
errors/missing-inputerrorsa missing input
command line3dAutomask -prefix mask.nii /nonexistent/x.nii
Rejected by both

AFNI exit 1: ** FATAL ERROR: Can’t open dataset ‘/nonexistent/x.nii’; the clean room exit 1: ** FATAL ERROR: Can’t open dataset ‘/nonexistent/x.nii’ (the same messages)

Rejected by both

AFNI exit 1: ** FATAL ERROR: Can’t open dataset ‘/nonexistent/x.nii’; the replica exit 1: ** FATAL ERROR: Can’t open dataset ‘/nonexistent/x.nii’ (the same messages)

4/4
divergences/default-prefixexpected divergencesno -prefix: AFNI writes its own format (automask+orig)
command line3dAutomask <generated: head-f32>
Expected divergence

AFNI’s default prefix ‘automask’ is its own dataset format, which the clean room does not write: it refuses (NIfTI only).

Expected divergence

AFNI’s default prefix ‘automask’ is its own dataset format, which the replica does not write: it refuses (NIfTI only).

2/2
global/existsglobal optionsan existing output (AFNI-11)
command line3dAutomask -prefix mask.nii <generated: head-f32> (with mask.nii present)
Bit-identical

Neither writes an output; the exit status (0) and the messages are AFNI’s. The default, fix, refuses instead (exit 1) and names AFNI-11 (the record’s list of what fix changed).

Bit-identical

Neither writes an output; the exit status (0) and the messages are AFNI’s. The default, fix, refuses instead (exit 1) and names AFNI-11 (the record’s list of what fix changed).

5/5
global/overwriteglobal options-overwrite
command line3dAutomask -overwrite -prefix mask.nii <generated: head-f32> (with mask.nii present)
Bit-identicalBit-identical7/7
global/deconflict-yesglobal optionsAFNI_DECONFLICT=YES: renamed
command lineAFNI_DECONFLICT=YES 3dAutomask -prefix mask.nii <generated: head-f32> (with mask.nii present)
Bit-identicalBit-identical7/7
global/D-deconflict-overwriteglobal options-DAFNI_DECONFLICT=OVERWRITE
command line3dAutomask -DAFNI_DECONFLICT=OVERWRITE -prefix mask.nii <generated: head-f32> (with mask.nii present)
Bit-identicalBit-identical7/7
global/D-write-type-2global options-DAFNI_NIFTI_WRITE_TYPE=2: NIfTI-2
command line3dAutomask -DAFNI_NIFTI_WRITE_TYPE=2 -prefix mask.nii <generated: head-f32>
Bit-identicalBit-identical7/7
environment/quiet-startupenvironment variablesAFNI_QUIET_STARTUP=YES: no author or output lines
command lineAFNI_QUIET_STARTUP=YES 3dAutomask -prefix mask.nii <generated: head-f32>
Bit-identicalBit-identical7/7
environment/message-prefixenvironment variablesAFNI_MESSAGE_PREFIX
command lineAFNI_MESSAGE_PREFIX=AM 3dAutomask -prefix mask.nii <generated: head-f32>
Bit-identicalBit-identical7/7
environment/echo-commandlineenvironment variablesAFNI_ECHO_COMMANDLINE=YES
command lineAFNI_ECHO_COMMANDLINE=YES 3dAutomask -prefix mask.nii <generated: head-f32>
Bit-identicalBit-identical7/7
environment/colorize-noenvironment variablesAFNI_MESSAGE_COLORIZE=NO with a warning
command lineAFNI_MESSAGE_COLORIZE=NO 3dAutomask -dilate 1 -erode 1 -prefix mask.nii <generated: head-f32>
Bit-identicalBit-identical7/7
environment/nifti-priority-qenvironment variablesAFNI_NIFTI_PRIORITY=Q on an oblique input
command lineAFNI_NIFTI_PRIORITY=Q 3dAutomask -prefix mask.nii <generated: head-oblique>
Bit-identicalBit-identical7/7
global/prefix-with-blankglobal optionsa -prefix with a blank (AFNI-12)
command line3dAutomask -prefix 'a b.nii' <generated: head-f32>
Rejected by both

AFNI refuses it (exit 1), and so does the clean room in reproduce, with AFNI’s messages; its default, fix, accepts it and names AFNI-12 (the record’s list of what fix changed).

Rejected by both

AFNI refuses it (exit 1), and so does the replica in reproduce, with AFNI’s messages; its default, fix, accepts it and names AFNI-12 (the record’s list of what fix changed).

5/5
global/input-dir-with-blankglobal optionsan input in a directory with a blank (AFNI-18)
command line3dAutomask -prefix mask.nii <generated: head-f32, in a directory with a blank>
Rejected by both

AFNI refuses it (exit 1), and so does the clean room in reproduce, with AFNI’s messages; its default, fix, accepts it and names AFNI-18 (the record’s list of what fix changed).

Rejected by both

AFNI refuses it (exit 1), and so does the replica in reproduce, with AFNI’s messages; its default, fix, accepts it and names AFNI-18 (the record’s list of what fix changed).

5/5
global/missing-directoryglobal optionsan output in a missing directory (AFNI-26)
command line3dAutomask -prefix nodir/mask.nii <generated: head-f32>
Bit-identical

Neither writes an output; the exit status (0) and the messages are AFNI’s. The default, fix, refuses instead (exit 1) and names AFNI-26 (the record’s list of what fix changed).

Bit-identical

Neither writes an output; the exit status (0) and the messages are AFNI’s. The default, fix, refuses instead (exit 1) and names AFNI-26 (the record’s list of what fix changed).

5/5
fmriprep/ds000005-task_mixedgamblestask_run_01fMRIPrep's callsfMRIPrep's call on the input of its BOLD reference's second skull strip (ds000005-task_mixedgamblestask_run_01)
command line3dAutomask -apply_prefix uni_xform_masked.nii.gz -dilate 1 -prefix uni_xform_mask.nii.gz <fMRIPrep ds000005-task_mixedgamblestask_run_01: unifized BOLD reference>
Bit-identical

AFNI at 1 and 6 threads and fMRIPrep 25.2.6’s AFNI give the bytes of the repository’s build; uni_xform_mask.nii.gz: fMRIPrep’s own output, the same bytes

Bit-identical

AFNI at 1 and 6 threads and fMRIPrep 25.2.6’s AFNI give the bytes of the repository’s build; uni_xform_mask.nii.gz: fMRIPrep’s own output, the same bytes

14/14
fmriprep/ds000005-task_mixedgamblestask_run_02fMRIPrep's callsfMRIPrep's call on the input of its BOLD reference's second skull strip (ds000005-task_mixedgamblestask_run_02)
command line3dAutomask -apply_prefix uni_xform_masked.nii.gz -dilate 1 -prefix uni_xform_mask.nii.gz <fMRIPrep ds000005-task_mixedgamblestask_run_02: unifized BOLD reference>
Bit-identical

AFNI at 1 and 6 threads and fMRIPrep 25.2.6’s AFNI give the bytes of the repository’s build; uni_xform_mask.nii.gz: fMRIPrep’s own output, the same bytes

Bit-identical

AFNI at 1 and 6 threads and fMRIPrep 25.2.6’s AFNI give the bytes of the repository’s build; uni_xform_mask.nii.gz: fMRIPrep’s own output, the same bytes

14/14
fmriprep/ds000005-task_mixedgamblestask_run_03fMRIPrep's callsfMRIPrep's call on the input of its BOLD reference's second skull strip (ds000005-task_mixedgamblestask_run_03)
command line3dAutomask -apply_prefix uni_xform_masked.nii.gz -dilate 1 -prefix uni_xform_mask.nii.gz <fMRIPrep ds000005-task_mixedgamblestask_run_03: unifized BOLD reference>
Bit-identical

AFNI at 1 and 6 threads and fMRIPrep 25.2.6’s AFNI give the bytes of the repository’s build; uni_xform_mask.nii.gz: fMRIPrep’s own output, the same bytes

Bit-identical

AFNI at 1 and 6 threads and fMRIPrep 25.2.6’s AFNI give the bytes of the repository’s build; uni_xform_mask.nii.gz: fMRIPrep’s own output, the same bytes

14/14
fmriprep/ds000228-task_pixarfMRIPrep's callsfMRIPrep's call on the input of its BOLD reference's second skull strip (ds000228-task_pixar)
command line3dAutomask -apply_prefix uni_xform_masked.nii.gz -dilate 1 -prefix uni_xform_mask.nii.gz <fMRIPrep ds000228-task_pixar: unifized BOLD reference>
Deliberate correction

messages, uni_xform_mask.nii.gz, uni_xform_masked.nii.gz: the data are AFNI’s, bit for bit; the header differs where AFNI reads the geometry from its own header extension of this AFNI-written input (7 significant digits; AFNI-10, which the clean room ignores, as 3dTshift’s does): AFNI run on the input without its extension writes the clean room’s header byte for byte; AFNI at 1 and 6 threads and fMRIPrep 25.2.6’s AFNI give the bytes of the repository’s build; uni_xform_mask.nii.gz: fMRIPrep’s own output, the same bytes

Bit-identical

AFNI at 1 and 6 threads and fMRIPrep 25.2.6’s AFNI give the bytes of the repository’s build; uni_xform_mask.nii.gz: fMRIPrep’s own output, the same bytes

14/14
fmriprep/ds000256-task_restbaselinefMRIPrep's callsfMRIPrep's call on the input of its BOLD reference's second skull strip (ds000256-task_restbaseline)
command line3dAutomask -apply_prefix uni_xform_masked.nii.gz -dilate 1 -prefix uni_xform_mask.nii.gz <fMRIPrep ds000256-task_restbaseline: unifized BOLD reference>
Deliberate correction

uni_xform_mask.nii.gz, uni_xform_masked.nii.gz: the data are AFNI’s, bit for bit; the header differs where AFNI reads the geometry from its own header extension of this AFNI-written input (7 significant digits; AFNI-10, which the clean room ignores, as 3dTshift’s does): AFNI run on the input without its extension writes the clean room’s header byte for byte; AFNI at 1 and 6 threads and fMRIPrep 25.2.6’s AFNI give the bytes of the repository’s build; uni_xform_mask.nii.gz: fMRIPrep’s own output, the same bytes

Bit-identical

AFNI at 1 and 6 threads and fMRIPrep 25.2.6’s AFNI give the bytes of the repository’s build; uni_xform_mask.nii.gz: fMRIPrep’s own output, the same bytes

14/14
fmriprep/ds000256-task_restfixationfixedfMRIPrep's callsfMRIPrep's call on the input of its BOLD reference's second skull strip (ds000256-task_restfixationfixed)
command line3dAutomask -apply_prefix uni_xform_masked.nii.gz -dilate 1 -prefix uni_xform_mask.nii.gz <fMRIPrep ds000256-task_restfixationfixed: unifized BOLD reference>
Deliberate correction

uni_xform_mask.nii.gz, uni_xform_masked.nii.gz: the data are AFNI’s, bit for bit; the header differs where AFNI reads the geometry from its own header extension of this AFNI-written input (7 significant digits; AFNI-10, which the clean room ignores, as 3dTshift’s does): AFNI run on the input without its extension writes the clean room’s header byte for byte; AFNI at 1 and 6 threads and fMRIPrep 25.2.6’s AFNI give the bytes of the repository’s build; uni_xform_mask.nii.gz: fMRIPrep’s own output, the same bytes

Bit-identical

AFNI at 1 and 6 threads and fMRIPrep 25.2.6’s AFNI give the bytes of the repository’s build; uni_xform_mask.nii.gz: fMRIPrep’s own output, the same bytes

14/14
fmriprep/ds000256-task_restfixationnofeedbackfMRIPrep's callsfMRIPrep's call on the input of its BOLD reference's second skull strip (ds000256-task_restfixationnofeedback)
command line3dAutomask -apply_prefix uni_xform_masked.nii.gz -dilate 1 -prefix uni_xform_mask.nii.gz <fMRIPrep ds000256-task_restfixationnofeedback: unifized BOLD reference>
Deliberate correction

uni_xform_mask.nii.gz, uni_xform_masked.nii.gz: the data are AFNI’s, bit for bit; the header differs where AFNI reads the geometry from its own header extension of this AFNI-written input (7 significant digits; AFNI-10, which the clean room ignores, as 3dTshift’s does): AFNI run on the input without its extension writes the clean room’s header byte for byte; AFNI at 1 and 6 threads and fMRIPrep 25.2.6’s AFNI give the bytes of the repository’s build; uni_xform_mask.nii.gz: fMRIPrep’s own output, the same bytes

Bit-identical

AFNI at 1 and 6 threads and fMRIPrep 25.2.6’s AFNI give the bytes of the repository’s build; uni_xform_mask.nii.gz: fMRIPrep’s own output, the same bytes

14/14
fmriprep/ds002422-task_arithm_run_01fMRIPrep's callsfMRIPrep's call on the input of its BOLD reference's second skull strip (ds002422-task_arithm_run_01)
command line3dAutomask -apply_prefix uni_xform_masked.nii.gz -dilate 1 -prefix uni_xform_mask.nii.gz <fMRIPrep ds002422-task_arithm_run_01: unifized BOLD reference>
Deliberate correction

messages, uni_xform_mask.nii.gz, uni_xform_masked.nii.gz: the data are AFNI’s, bit for bit; the header differs where AFNI reads the geometry from its own header extension of this AFNI-written input (7 significant digits; AFNI-10, which the clean room ignores, as 3dTshift’s does): AFNI run on the input without its extension writes the clean room’s header byte for byte; AFNI at 1 and 6 threads and fMRIPrep 25.2.6’s AFNI give the bytes of the repository’s build; uni_xform_mask.nii.gz: fMRIPrep’s own output, the same bytes

Bit-identical

AFNI at 1 and 6 threads and fMRIPrep 25.2.6’s AFNI give the bytes of the repository’s build; uni_xform_mask.nii.gz: fMRIPrep’s own output, the same bytes

14/14
fmriprep/ds002422-task_arithm_run_02fMRIPrep's callsfMRIPrep's call on the input of its BOLD reference's second skull strip (ds002422-task_arithm_run_02)
command line3dAutomask -apply_prefix uni_xform_masked.nii.gz -dilate 1 -prefix uni_xform_mask.nii.gz <fMRIPrep ds002422-task_arithm_run_02: unifized BOLD reference>
Deliberate correction

messages, uni_xform_mask.nii.gz, uni_xform_masked.nii.gz: the data are AFNI’s, bit for bit; the header differs where AFNI reads the geometry from its own header extension of this AFNI-written input (7 significant digits; AFNI-10, which the clean room ignores, as 3dTshift’s does): AFNI run on the input without its extension writes the clean room’s header byte for byte; AFNI at 1 and 6 threads and fMRIPrep 25.2.6’s AFNI give the bytes of the repository’s build; uni_xform_mask.nii.gz: fMRIPrep’s own output, the same bytes

Bit-identical

AFNI at 1 and 6 threads and fMRIPrep 25.2.6’s AFNI give the bytes of the repository’s build; uni_xform_mask.nii.gz: fMRIPrep’s own output, the same bytes

14/14
fmriprep/ds002422-task_restfMRIPrep's callsfMRIPrep's call on the input of its BOLD reference's second skull strip (ds002422-task_rest)
command line3dAutomask -apply_prefix uni_xform_masked.nii.gz -dilate 1 -prefix uni_xform_mask.nii.gz <fMRIPrep ds002422-task_rest: unifized BOLD reference>
Deliberate correction

uni_xform_mask.nii.gz, uni_xform_masked.nii.gz: the data are AFNI’s, bit for bit; the header differs where AFNI reads the geometry from its own header extension of this AFNI-written input (7 significant digits; AFNI-10, which the clean room ignores, as 3dTshift’s does): AFNI run on the input without its extension writes the clean room’s header byte for byte; AFNI at 1 and 6 threads and fMRIPrep 25.2.6’s AFNI give the bytes of the repository’s build; uni_xform_mask.nii.gz: fMRIPrep’s own output, the same bytes

Bit-identical

AFNI at 1 and 6 threads and fMRIPrep 25.2.6’s AFNI give the bytes of the repository’s build; uni_xform_mask.nii.gz: fMRIPrep’s own output, the same bytes

14/14
fmriprep/ds002603-task_wm_acq_Mb4Mm3Tr550fMRIPrep's callsfMRIPrep's call on the input of its BOLD reference's second skull strip (ds002603-task_wm_acq_Mb4Mm3Tr550)
command line3dAutomask -apply_prefix uni_xform_masked.nii.gz -dilate 1 -prefix uni_xform_mask.nii.gz <fMRIPrep ds002603-task_wm_acq_Mb4Mm3Tr550: unifized BOLD reference>
Deliberate correction

uni_xform_mask.nii.gz, uni_xform_masked.nii.gz: the data are AFNI’s, bit for bit; the header differs where AFNI reads the geometry from its own header extension of this AFNI-written input (7 significant digits; AFNI-10, which the clean room ignores, as 3dTshift’s does): AFNI run on the input without its extension writes the clean room’s header byte for byte; AFNI at 1 and 6 threads and fMRIPrep 25.2.6’s AFNI give the bytes of the repository’s build; uni_xform_mask.nii.gz: fMRIPrep’s own output, the same bytes

Bit-identical

AFNI at 1 and 6 threads and fMRIPrep 25.2.6’s AFNI give the bytes of the repository’s build; uni_xform_mask.nii.gz: fMRIPrep’s own output, the same bytes

14/14
fmriprep/ds002603-task_wm_acq_SeqMm3Tr2000fMRIPrep's callsfMRIPrep's call on the input of its BOLD reference's second skull strip (ds002603-task_wm_acq_SeqMm3Tr2000)
command line3dAutomask -apply_prefix uni_xform_masked.nii.gz -dilate 1 -prefix uni_xform_mask.nii.gz <fMRIPrep ds002603-task_wm_acq_SeqMm3Tr2000: unifized BOLD reference>
Deliberate correction

uni_xform_mask.nii.gz, uni_xform_masked.nii.gz: the data are AFNI’s, bit for bit; the header differs where AFNI reads the geometry from its own header extension of this AFNI-written input (7 significant digits; AFNI-10, which the clean room ignores, as 3dTshift’s does): AFNI run on the input without its extension writes the clean room’s header byte for byte; AFNI at 1 and 6 threads and fMRIPrep 25.2.6’s AFNI give the bytes of the repository’s build; uni_xform_mask.nii.gz: fMRIPrep’s own output, the same bytes

Bit-identical

AFNI at 1 and 6 threads and fMRIPrep 25.2.6’s AFNI give the bytes of the repository’s build; uni_xform_mask.nii.gz: fMRIPrep’s own output, the same bytes

14/14
fmriprep/ds003592-ses_1_task_rest_echo_1fMRIPrep's callsfMRIPrep's call on the input of its BOLD reference's second skull strip (ds003592-ses_1_task_rest_echo_1)
command line3dAutomask -apply_prefix uni_xform_masked.nii.gz -dilate 1 -prefix uni_xform_mask.nii.gz <fMRIPrep ds003592-ses_1_task_rest_echo_1: unifized BOLD reference>
Deliberate correction

messages, uni_xform_mask.nii.gz, uni_xform_masked.nii.gz: the data are AFNI’s, bit for bit; the header differs where AFNI reads the geometry from its own header extension of this AFNI-written input (7 significant digits; AFNI-10, which the clean room ignores, as 3dTshift’s does): AFNI run on the input without its extension writes the clean room’s header byte for byte; AFNI at 1 and 6 threads and fMRIPrep 25.2.6’s AFNI give the bytes of the repository’s build; uni_xform_mask.nii.gz: fMRIPrep’s own output, the same bytes

Bit-identical

AFNI at 1 and 6 threads and fMRIPrep 25.2.6’s AFNI give the bytes of the repository’s build; uni_xform_mask.nii.gz: fMRIPrep’s own output, the same bytes

14/14
fmriprep/ds004884-ses_217_task_naming40_acq_epfid2_dir_AP_run_2fMRIPrep's callsfMRIPrep's call on the input of its BOLD reference's second skull strip (ds004884-ses_217_task_naming40_acq_epfid2_dir_AP_run_2)
command line3dAutomask -apply_prefix uni_xform_masked.nii.gz -dilate 1 -prefix uni_xform_mask.nii.gz <fMRIPrep ds004884-ses_217_task_naming40_acq_epfid2_dir_AP_run_2: unifized BOLD reference>
Deliberate correction

uni_xform_mask.nii.gz, uni_xform_masked.nii.gz: the data are AFNI’s, bit for bit; the header differs where AFNI reads the geometry from its own header extension of this AFNI-written input (7 significant digits; AFNI-10, which the clean room ignores, as 3dTshift’s does): AFNI run on the input without its extension writes the clean room’s header byte for byte; AFNI at 1 and 6 threads and fMRIPrep 25.2.6’s AFNI give the bytes of the repository’s build; uni_xform_mask.nii.gz: fMRIPrep’s own output, the same bytes

Bit-identical

AFNI at 1 and 6 threads and fMRIPrep 25.2.6’s AFNI give the bytes of the repository’s build; uni_xform_mask.nii.gz: fMRIPrep’s own output, the same bytes

14/14
fmriprep/ds004884-ses_217_task_rest_acq_epfidp2p2_dir_AP_run_9fMRIPrep's callsfMRIPrep's call on the input of its BOLD reference's second skull strip (ds004884-ses_217_task_rest_acq_epfidp2p2_dir_AP_run_9)
command line3dAutomask -apply_prefix uni_xform_masked.nii.gz -dilate 1 -prefix uni_xform_mask.nii.gz <fMRIPrep ds004884-ses_217_task_rest_acq_epfidp2p2_dir_AP_run_9: unifized BOLD reference>
Deliberate correction

uni_xform_mask.nii.gz, uni_xform_masked.nii.gz: the data are AFNI’s, bit for bit; the header differs where AFNI reads the geometry from its own header extension of this AFNI-written input (7 significant digits; AFNI-10, which the clean room ignores, as 3dTshift’s does): AFNI run on the input without its extension writes the clean room’s header byte for byte; AFNI at 1 and 6 threads and fMRIPrep 25.2.6’s AFNI give the bytes of the repository’s build; uni_xform_mask.nii.gz: fMRIPrep’s own output, the same bytes

Bit-identical

AFNI at 1 and 6 threads and fMRIPrep 25.2.6’s AFNI give the bytes of the repository’s build; uni_xform_mask.nii.gz: fMRIPrep’s own output, the same bytes

14/14
fmriprep/ds004958-task_restfMRIPrep's callsfMRIPrep's call on the input of its BOLD reference's second skull strip (ds004958-task_rest)
command line3dAutomask -apply_prefix uni_xform_masked.nii.gz -dilate 1 -prefix uni_xform_mask.nii.gz <fMRIPrep ds004958-task_rest: unifized BOLD reference>
Deliberate correction

messages, uni_xform_mask.nii.gz, uni_xform_masked.nii.gz: the data are AFNI’s, bit for bit; the header differs where AFNI reads the geometry from its own header extension of this AFNI-written input (7 significant digits; AFNI-10, which the clean room ignores, as 3dTshift’s does): AFNI run on the input without its extension writes the clean room’s header byte for byte; AFNI at 1 and 6 threads and fMRIPrep 25.2.6’s AFNI give the bytes of the repository’s build; uni_xform_mask.nii.gz: fMRIPrep’s own output, the same bytes

Bit-identical

AFNI at 1 and 6 threads and fMRIPrep 25.2.6’s AFNI give the bytes of the repository’s build; uni_xform_mask.nii.gz: fMRIPrep’s own output, the same bytes

14/14
fmriprep/ds005003-ses_1_task_motorfMRIPrep's callsfMRIPrep's call on the input of its BOLD reference's second skull strip (ds005003-ses_1_task_motor)
command line3dAutomask -apply_prefix uni_xform_masked.nii.gz -dilate 1 -prefix uni_xform_mask.nii.gz <fMRIPrep ds005003-ses_1_task_motor: unifized BOLD reference>
Bit-identical

AFNI at 1 and 6 threads and fMRIPrep 25.2.6’s AFNI give the bytes of the repository’s build; uni_xform_mask.nii.gz: fMRIPrep’s own output, the same bytes

Bit-identical

AFNI at 1 and 6 threads and fMRIPrep 25.2.6’s AFNI give the bytes of the repository’s build; uni_xform_mask.nii.gz: fMRIPrep’s own output, the same bytes

14/14
fmriprep/ds005003-ses_1_task_restfMRIPrep's callsfMRIPrep's call on the input of its BOLD reference's second skull strip (ds005003-ses_1_task_rest)
command line3dAutomask -apply_prefix uni_xform_masked.nii.gz -dilate 1 -prefix uni_xform_mask.nii.gz <fMRIPrep ds005003-ses_1_task_rest: unifized BOLD reference>
Bit-identical

AFNI at 1 and 6 threads and fMRIPrep 25.2.6’s AFNI give the bytes of the repository’s build; uni_xform_mask.nii.gz: fMRIPrep’s own output, the same bytes

Bit-identical

AFNI at 1 and 6 threads and fMRIPrep 25.2.6’s AFNI give the bytes of the repository’s build; uni_xform_mask.nii.gz: fMRIPrep’s own output, the same bytes

14/14
fmriprep/ds005040-task_clip_run_1 worked examplefMRIPrep's callsfMRIPrep's call on the input of its BOLD reference's second skull strip (ds005040-task_clip_run_1)
command line3dAutomask -apply_prefix uni_xform_masked.nii.gz -dilate 1 -prefix uni_xform_mask.nii.gz <fMRIPrep ds005040-task_clip_run_1: unifized BOLD reference>
Bit-identical

AFNI at 1 and 6 threads and fMRIPrep 25.2.6’s AFNI give the bytes of the repository’s build; uni_xform_mask.nii.gz: fMRIPrep’s own output, the same bytes

Bit-identical

AFNI at 1 and 6 threads and fMRIPrep 25.2.6’s AFNI give the bytes of the repository’s build; uni_xform_mask.nii.gz: fMRIPrep’s own output, the same bytes

14/14
fmriprep/ds005454-task_restfMRIPrep's callsfMRIPrep's call on the input of its BOLD reference's second skull strip (ds005454-task_rest)
command line3dAutomask -apply_prefix uni_xform_masked.nii.gz -dilate 1 -prefix uni_xform_mask.nii.gz <fMRIPrep ds005454-task_rest: unifized BOLD reference>
Deliberate correction

uni_xform_mask.nii.gz, uni_xform_masked.nii.gz: the data are AFNI’s, bit for bit; the header differs where AFNI reads the geometry from its own header extension of this AFNI-written input (7 significant digits; AFNI-10, which the clean room ignores, as 3dTshift’s does): AFNI run on the input without its extension writes the clean room’s header byte for byte; AFNI at 1 and 6 threads and fMRIPrep 25.2.6’s AFNI give the bytes of the repository’s build; uni_xform_mask.nii.gz: fMRIPrep’s own output, the same bytes

Bit-identical

AFNI at 1 and 6 threads and fMRIPrep 25.2.6’s AFNI give the bytes of the repository’s build; uni_xform_mask.nii.gz: fMRIPrep’s own output, the same bytes

14/14
fmriprep/ds005752-ses_01_task_rest_dir_forwardfMRIPrep's callsfMRIPrep's call on the input of its BOLD reference's second skull strip (ds005752-ses_01_task_rest_dir_forward)
command line3dAutomask -apply_prefix uni_xform_masked.nii.gz -dilate 1 -prefix uni_xform_mask.nii.gz <fMRIPrep ds005752-ses_01_task_rest_dir_forward: unifized BOLD reference>
Bit-identical

AFNI at 1 and 6 threads and fMRIPrep 25.2.6’s AFNI give the bytes of the repository’s build; uni_xform_mask.nii.gz: fMRIPrep’s own output, the same bytes

Bit-identical

AFNI at 1 and 6 threads and fMRIPrep 25.2.6’s AFNI give the bytes of the repository’s build; uni_xform_mask.nii.gz: fMRIPrep’s own output, the same bytes

14/14
fmriprep/ds005752-ses_01_task_rest_dir_reversefMRIPrep's callsfMRIPrep's call on the input of its BOLD reference's second skull strip (ds005752-ses_01_task_rest_dir_reverse)
command line3dAutomask -apply_prefix uni_xform_masked.nii.gz -dilate 1 -prefix uni_xform_mask.nii.gz <fMRIPrep ds005752-ses_01_task_rest_dir_reverse: unifized BOLD reference>
Bit-identical

AFNI at 1 and 6 threads and fMRIPrep 25.2.6’s AFNI give the bytes of the repository’s build; uni_xform_mask.nii.gz: fMRIPrep’s own output, the same bytes

Bit-identical

AFNI at 1 and 6 threads and fMRIPrep 25.2.6’s AFNI give the bytes of the repository’s build; uni_xform_mask.nii.gz: fMRIPrep’s own output, the same bytes

14/14
real/sub-01_task-mixedgamblestask_run-01real imagesAFNI's default on a BOLD series (sub-01_task-mixedgamblestask_run-01)
command line3dAutomask -prefix mask.nii.gz ds000005/sub-01/func/sub-01_task-mixedgamblestask_run-01_bold.nii.gz
Bit-identical

AFNI at 1 thread and fMRIPrep 25.2.6’s AFNI give the bytes of the repository’s build

Bit-identical

AFNI at 1 thread and fMRIPrep 25.2.6’s AFNI give the bytes of the repository’s build

9/9
real/sub-1_task-rest_acq-v4real imagesAFNI's default on a BOLD series (sub-1_task-rest_acq-v4)
command line3dAutomask -prefix mask.nii.gz ds001600/sub-1/func/sub-1_task-rest_acq-v4_bold.nii.gz
Bit-identical

AFNI at 1 thread and fMRIPrep 25.2.6’s AFNI give the bytes of the repository’s build

Bit-identical

AFNI at 1 thread and fMRIPrep 25.2.6’s AFNI give the bytes of the repository’s build

9/9
real/t1w-brainreal imagesthe skull-stripped T1w of an fMRIPrep run (ds000005, sMRIPrep's desc-brain)
command line3dAutomask -prefix mask.nii.gz <ds000005: sMRIPrep's T1w × its brain mask>
Bit-identical

AFNI at 1 thread and fMRIPrep 25.2.6’s AFNI give the bytes of the repository’s build

Bit-identical

AFNI at 1 thread and fMRIPrep 25.2.6’s AFNI give the bytes of the repository’s build

9/9
difftest/float64-negative-zerorandom testingBit-identical–9/9
difftest/oblique-orientation-lettersrandom testingBit-identical–7/7
difftest/peels-0-clustering-messagesrandom testingBit-identical–7/7
difftest/plane-oblique-si-orientationrandom testingBit-identical–7/7
difftest/slice-code-2-masked-noticesrandom testingBit-identical–9/9
difftest/tiny-oblique-quaternion-residuerandom testingBit-identical–7/7
difftest/slice-code-2-masked-afni26random testing–Bit-identical9/9

Speed

Speed, both ports against the original: AFNI 3dAutomask 25.2.09

Seconds for the whole program, informal timings of 2026-10-24.

  • AFNI 3dAutomask 25.2.09, 1 thread
  • replica, 1 thread
  • replica, 6 threads
  • clean room, 1 thread
  • clean room, 6 threads
  1. fMRIPrep's call, ds005454 BOLD referenceAFNI 3dAutomask 25.2.09, 1 thread 1.63 sreplica, 1 thread 1.85 sreplica, 6 threads 1.91 sclean room, 1 thread 1.66 sclean room, 6 threads 0.72 s0.98×
  2. a BOLD series, ds005040AFNI 3dAutomask 25.2.09, 1 thread 0.39 sreplica, 1 thread 0.25 sreplica, 6 threads 0.24 sclean room, 1 thread 0.22 sclean room, 6 threads 0.19 s1.8×
0.1 s1 s10 s

Each program run once (the replica after one warm-up run), reading and writing compressed NIfTI included, on a 12-CPU machine shared with another job throughout (load 17 to 22): indications, not a benchmark report. The ratio is AFNI’s time over the clean room’s, at one thread each. Tool: validation/investigations/afni-masks-original/timing.py.

3dAutomask is quick in every port: its time is mostly reading and writing compressed NIfTI, and AFNI’s is single-threaded. These are informal timings on a shared machine (findings); there is no benchmark report yet.

Corrections by default

Where AFNI has a deterministic defect that changes a result, both ports correct it by default, fix, and warn. upstream_defects="reproduce" (on the command line --upstream-defects reproduce from the next release, LARMORX_UPSTREAM_DEFECTS=reproduce in 0.1.0rc1) gives AFNI’s behaviour, and the records compare with it. The records’ lists of what fix changed meet AFNI-43 (-peels 0), AFNI-01 (a negative scale factor dropped while averaging a series) and the dataset defects 3dAutomask shares with 3dTshift (AFNI-02, AFNI-07, AFNI-11, AFNI-12, AFNI-16, AFNI-18, AFNI-23, AFNI-26) (register). AFNI-41 (-nbhrs outside 6 to 26, silently 17 in AFNI) is refused by default before any input is read, in the same words by both ports (decided 2026-10-23), with reproduce giving AFNI’s mask. AFNI-42 (-NN1/2/3 reset the peel threshold) and AFNI-44 (the large-hole filling decided by the number of slices) are kept as AFNI’s and documented.

Refused by default: AFNI-41. -nbhrs takes a value between 6 and 26, as the help says; outside that range AFNI accepts it silently and peels as with 17. A value outside the range makes no sense, and the project’s rule for that case (2026-10-23) is a clear error: by default the replica stops with a message naming AFNI-41, before reading the input; in reproduce it gives AFNI’s mask, bit for bit. The case explorer shows it: options/nbhrs-40, and in the standard comparison nbhrs-5, nbhrs-27, nbhrs-0 and nbhrs-40-then-NN1, each with the verdict deliberate correction (refused by default) and bit-identical to AFNI in reproduce.

Random testing

Random testing runs random valid inputs and option combinations through the original and both ports, in long runs of several hours: inputs from seeded plain arithmetic (up to 32³ voxels; orientations, data types, scale factors and value ranges of every kind), with options drawn in random combinations from those the program accepts (how random testing works). A difference is shrunk to the smallest input and fewest options that still show it, then fixed and kept as a case of the record, registered as a defect of the original, or reported. The latest long run:

Random testing, both ports against the original: AFNI 3dAutomask 26.3.05, the long run of 2026-10-25 (seed 20261025, 6.2 h for 21 programs): random valid inputs and option combinations, each run through all three

ProgramRandom casesExtremeOptions coveredAgreeglibcNoticesUndefinedDifferHeavyGuardUnclear
3dAutomaskreplica, clean room1,8890171,835·2·52···
  • fixed in LARMORX 3dAutomask (clean room), 6 case(s): the fix pass: AFNI-43 named where only the clustering messages differ and are shown (not with -q: the random-testing run of 2026-10-29, afni-automask:20261029:3425), in the replica’s words; a negative -peels peels once in both modes, as the replica’s default (2026-10-29) (details)
  • fixed in LARMORX 3dAutomask (clean room), 11 case(s): the fix pass, the notices in fix and in reproduce: AFNI-03 named in both modes and AFNI-08 in reproduce, in the replica’s words; the orientation letters by the NIfTI library’s closest-orientation method, in double (observed on 1,604 matrices; -SI follows) (2026-10-29) (details)
  • fixed in LARMORX 3dAutomask (clean room), 1 case(s): mask.nii.gz: data, messages: AFNI's, byte for byte: the orientation letters by the NIfTI library’s closest-orientation method, in double (observed on 1,604 matrices; -SI follows) (2026-10-29) (details)
  • fixed in LARMORX 3dAutomask (clean room), 1 case(s): mask.nii.gz: header bytes: the quaternion’s polar decomposition: AFNI takes 1/det in x87 extended precision (observed on 1,200 matrices through NIfTI-2 outputs); fixed in the clean room (2026-10-29) (details)
  • fixed in LARMORX 3dAutomask (clean room), 4 case(s): mask.nii: header bytes: the quaternion’s polar decomposition: AFNI takes 1/det in x87 extended precision (observed on 1,200 matrices through NIfTI-2 outputs); fixed in the clean room (2026-10-29) (details)
  • fixed in LARMORX 3dAutomask (clean room), 1 case(s): masked.nii: data: int8, uint16, int32, uint32 and float64 data without a usable slope are only converted, so a float64 -0 stays -0 (2026-10-29) (details)
  • fixed in LARMORX 3dAutomask (clean room), 28 case(s): messages: AFNI's, byte for byte: the orientation letters by the NIfTI library’s closest-orientation method, in double (observed on 1,604 matrices; -SI follows) (2026-10-29) (details)
  • fixed in LARMORX 3dAutomask (replica), 6 case(s): the fix pass: -peels 0 on a 4D image: the default (no peeling, as the help says) gives the same mask as AFNI’s single peel but AFNI’s clustering messages count other voxels, and the replica named AFNI-43 only where the masks differ; now also where only the messages do

Notices: a correction named in a case where it changed nothing the input shows (defect, cases): 3dAutomask: AFNI-01 (2), AFNI-08 (13).

Every case is drawn from its tool, the run's seed and its number, and replays anywhere: how random testing works; each difference's triage: validation/difftest/triage.toml.

Known limits

LARMORX writes NIfTI only: AFNI’s default prefix automask is its own dataset format, which both ports refuse (an expected divergence). The clean room ignores AFNI’s own header extension in an AFNI-written input (AFNI-10), as 3dTshift’s clean room does; the replica reads it as AFNI does. The cross-platform checks (how they work) hold the other five platforms to LARMORX’s own Linux output.

History

  • 2026-10-23: AFNI-41 refused by default, by the project’s rule for an input that makes no sense.
  • 2026-10-29: AFNI 26.3.05 releases the Medical College of Wisconsin’s code under CC BY 4.0; the replica moves from larmorx-gpl into larmorx, and both ports follow AFNI 26.3.05.
  • 2026-10-30: both records made again against AFNI 26.3.05: 130 and 125 cases.

Data credits

The data this page shows or names: each dataset with its version, DOI and licence, from LARMORX's test-data catalog. Public-domain dedications (CC0, PDDL) ask nothing; the datasets are cited all the same.