Files
2025-02-BondingArtifact/.specstory/history/2025-11-01_15-48Z-png-rendering-error-troubleshooting.md

4.1 KiB
Raw Blame History

PNG rendering error troubleshooting (2025-11-01 15:48Z)

User (2025-11-01 15:48Z)

why is this happening to render

libpng error: IDAT: CRC error

ImageInput::read_image() failed: PNG read error: IDAT: CRC error

PNG library error

IMB_load_image_from_memory: unknown file-format (R:\Creative\artsy\maya\2025-02 BondingArtifact\Renders\3_background\2025-10-26\3_background_dark\3_background_dark_5489.png)


Assistant (default)


Reviewing the open file to understand the render configuration and identify why PNGs are being corrupted.


Tool use: read_file

Read file: c:\Program Files\Blender Foundation\Flamenco 3.7\scripts\TalkingHeads cycles_optix_gpu_NOBAT.js


Tool use: codebase_search

Searched codebase "How are PNG images saved or written during rendering?" • **3** results
File Lines
.gitignore L134
.gitattributes L47
misc/checklist.txt L52


Checking the render configuration and the corrupt file path for clues.


Tool use: codebase_search

Searched codebase "Where are render output files written or saved?" • **10** results
File Lines
.gitignore L134
assets/Objects/gems/tx/BlackOnyx/Designer/BlackOnyx.sbs L3573-3642
misc/checklist.txt L52
misc/metadata.bat L23
misc/filenames.bat L1
.cursorindexingignore L3
.gitattributes L47
assets/Objects/gems/tx/BlackOnyx/Designer/BlackOnyx.sbs L1475-1533
misc/BRANDING.txt L2
assets/Objects/gems/tx/BlackOnyx/Designer/BlackOnyx.sbs L1976-2034

Tool use: grep

Grep for "3_background" • 0 matches

Output mode: content

No matches found



This PNG CRC error indicates the file was corrupted during render. The path R:\... suggests a network drive, which is a common source.

Common causes:

  • Network storage issues: timeouts, interruptions, or bad connections corrupt writes
  • Concurrent access: another process (comp, preview, antivirus) reading while a worker writes
  • Incomplete writes: file read before the write finishes
  • Antivirus/backup: scanning or copying files while theyre being written

Checking the render script for timing or synchronization issues:


Tool use: read_file

Read file: c:\Program Files\Blender Foundation\Flamenco 3.7\scripts\TalkingHeads cycles_optix_gpu_NOBAT.js



Solutions to fix this:

Immediate fixes:

  1. Check if the file is actually corrupted

    • Verify frame 5489 on the R: drive
    • If corrupted, re-render just that frame
  2. Render to local storage first, then copy

    • Change the render output to a local drive (C: or D:)
    • After rendering completes, copy to the network drive
    • Reduces corruption risk from network interruptions
  3. Increase chunk_size

    • In your Flamenco job settings, increase chunk_size (currently defaults to 1)
    • Fewer concurrent writes = less contention
  4. Temporarily disable antivirus/backup scanning

    • Exclude the render output folder from real-time scanning
    • Network drives are often scanned aggressively

Network storage specific:

  1. Check network drive health

    • Test the R: drive connection stability
    • Run chkdsk R: /f if it's accessible
    • Check for network timeouts or packet loss
  2. Add file write delay/sync

    • If possible, configure Blender to flush writes completely before finishing
    • Use --python-expr to add a small delay after rendering completes

For the current corrupted frame:

The file 3_background_dark_5489.png is corrupted. Either:

  • Re-render frame 5489 specifically, or
  • Skip it and continue (if it's not critical)

Should I help modify the Flamenco script to render to a local directory first, or focus on diagnosing the network storage issue?