#15020
Add native Hunyuan3D 2.1 PBR paint (mesh -> multiview -> textured GLB)
Comfy-Org/ComfyUI Updated
03 / PUBLIC PULL REQUESTS
Work beyond
the node.
Upstream proposals, improvements in my ComfyUI fork, and the pull requests behind Cloud Offload and UniMate.
1 upstream PR7 fork PRs123 project PRsStatus checked 2026-09-30
No pull requests match this status.
Upstream contributions
Changes submitted to the wider ecosystem. An open PR is proposed work, not a released core feature.
Open
ComfyUI fork work
Development and historical pull requests in my own ComfyUI fork. “Merged in fork” does not mean merged upstream.
#2
Hunyuan3D 2.1 texture-paint subsystem (feature-only rebase of hunyuan21-paint)
jethac/ComfyUI Updated
#11
Add live ComfyUI runner identity
jethac/ComfyUI Updated
#10
Harden mesh export batch and GLB attribute handling
jethac/ComfyUI Updated
#9
test(trellis2): cover quantized projection attention
jethac/ComfyUI Updated
#3
Trellis2/Pixal3D support with quantized/GGUF-safe weight handling
jethac/ComfyUI Updated
#4
Clean-room rewrite of Hunyuan3D 2.1 paint tainted surfaces
jethac/ComfyUI Updated
#1
Trellis2/Pixal3D support with quantized/GGUF-safe weight handling
jethac/ComfyUI Updated
Project development
Public pull requests across Cloud Offload, its service, and UniMate, including collaborator contributions.
Browse 123 project pull requests
#1
Implement UniMate animation nodes with portable cloud values
jethac/ComfyUI-UniMate Updated
#114
Run tests on the self-hosted runner
jethac/cloud-offload Updated
#113
Support native paint offload and Japan H100 release qualification
jethac/cloud-offload Updated
#14
Transport native meshes and paint cameras across cloud partitions
jethac/ComfyUI-Cloud-Offload Updated
#112
fix: bind prepared provider identity and redact diagnostics
jethac/cloud-offload Updated
#111
fix: bind plan candidates to current storage
jethac/cloud-offload Updated
#110
feat: add cloud plan protocol foundation
jethac/cloud-offload Updated
#109
Clamp M7 startup waits to absolute deadline
jethac/cloud-offload Updated
#108
Harden M7 startup failure handling
jethac/cloud-offload Updated
#107
Fix isolated storage policy runtime updates
jethac/cloud-offload Updated
#106
Fix M7 restart authentication state
jethac/cloud-offload Updated
#105
Fix isolated M7 artifact bootstrap and startup proof
jethac/cloud-offload Updated
#104
Pin M7 worker image digest
jethac/cloud-offload Updated
#103
Harden deterministic M7 worker image contexts
jethac/cloud-offload Updated
#99
Fix runtime defects surfaced by the M7 canary campaign
jethac/cloud-offload Updated
#11
Wire up the web test suite, dev requirements, and CI for both halves
jethac/ComfyUI-Cloud-Offload Updated
#10
Raise managed RunPod volume minimum to the v2 10 GB floor
jethac/ComfyUI-Cloud-Offload Updated
#98
Migrate datetime.utcnow() to a timezone-aware UTC clock
jethac/cloud-offload Updated
#97
Fix dispatcher config refresh + multipart upload race; add CI
jethac/cloud-offload Updated
#96
Exercise M7 release controller offline and on real pods; fix six gate/runtime bugs; add operator runbook
jethac/cloud-offload Updated
#95
Retry transient RunPod S3 GET signature errors
jethac/cloud-offload Updated
#94
Exclude local worktrees from runner images
jethac/cloud-offload Updated
#93
Fix node-pack installs to preserve pinned PyTorch
jethac/cloud-offload Updated
#92
Use configured scratch for manifest refresh
jethac/cloud-offload Updated
#91
Retry transient RunPod S3 transport failures
jethac/cloud-offload Updated
#90
Bound stale multipart retries
jethac/cloud-offload Updated
#89
Fix regional replication scratch placement
jethac/cloud-offload Updated
#87
Migrate the RunPod connector to REST API v2
jethac/cloud-offload Updated
#86
Resume multipart uploads around partial part records
jethac/cloud-offload Updated
#85
Match RunPod large-file multipart size
jethac/cloud-offload Updated
#84
Extend RunPod gateway retry budget
jethac/cloud-offload Updated
#83
Fall back when RunPod multipart listing is unavailable
jethac/cloud-offload Updated
#82
Record safe S3 replication failure routing
jethac/cloud-offload Updated
#81
Restore stable RunPod upload pressure
jethac/cloud-offload Updated
#80
Increase resumable S3 upload throughput
jethac/cloud-offload Updated
#79
Reduce RunPod range request pressure
jethac/cloud-offload Updated
#78
Prefer the fullest multipart session
jethac/cloud-offload Updated
#77
Add safe per-job result routes
jethac/cloud-offload Updated
#76
Bound RunPod multipart part waits
jethac/cloud-offload Updated
#75
Bound RunPod range request waits
jethac/cloud-offload Updated
#74
Use multipart uploads for medium RunPod objects
jethac/cloud-offload Updated
#73
Recover completed RunPod multipart merges
jethac/cloud-offload Updated
#72
Refresh prepared manifests for current profiles
jethac/cloud-offload Updated
#71
Redact replication failure records
jethac/cloud-offload Updated
#70
Resume RunPod multipart uploads
jethac/cloud-offload Updated
#69
Retry RunPod S3 gateway 5xx responses
jethac/cloud-offload Updated
#68
Resume partial prepared-state ranges
jethac/cloud-offload Updated
#67
Retry incomplete prepared-state range streams
jethac/cloud-offload Updated
#66
Limit prepared-state range body waits
jethac/cloud-offload Updated
#65
Speed up prepared-state S3 downloads
jethac/cloud-offload Updated
#64
Resume regional copies at missing objects
jethac/cloud-offload Updated
#63
Retry RunPod 524 range reads
jethac/cloud-offload Updated
#62
Harden RunPod large multipart uploads
jethac/cloud-offload Updated
#61
Fix large S3 prepared-state verification
jethac/cloud-offload Updated
#60
Harden long prepared-state S3 transfers
jethac/cloud-offload Updated
#59
Bind M7 corruption canary to the release region
jethac/cloud-offload Updated
#58
Add M7 continuous release controller
jethac/cloud-offload Updated
#57
Record M6 regional replication production evidence
jethac/cloud-offload Updated
#56
Stream verified RunPod S3 downloads with GET
jethac/cloud-offload Updated
#55
Measure runtime bytes for regional demand
jethac/cloud-offload Updated
#54
Fix cold preflight in requested regions
jethac/cloud-offload Updated
#53
Validate M6 shadow recommendations on real followup
jethac/cloud-offload Updated
#52
Complete automatic M6 regional replication controls
jethac/cloud-offload Updated
#51
Add budgeted M6 regional replica controller
jethac/cloud-offload Updated
#50
Add M6 regional replication shadow recommendations
jethac/cloud-offload Updated
#49
Record M5 workflow capsule production proof
jethac/cloud-offload Updated
#48
Keep boot report setting after cached image layers
jethac/cloud-offload Updated
#47
Attribute pre-ComfyUI cache restores
jethac/cloud-offload Updated
#46
Declare pinned runner runtime identity
jethac/cloud-offload Updated
#45
Fix RunPod prepared-cache namespace
jethac/cloud-offload Updated
#44
Support explicit runner image profile aliases
jethac/cloud-offload Updated
#43
Split runner image into parallel pull layers
jethac/cloud-offload Updated
#42
Preserve the first-runner lease window after Pod binding
jethac/cloud-offload Updated
#41
Filter RunPod offers by prepared-storage region stock
jethac/cloud-offload Updated
#40
feat: prepare reproducible runtime bundles
jethac/cloud-offload Updated
#39
Add prepared workflow capsules
jethac/cloud-offload Updated
#9
Use workflow capsule preflight
jethac/ComfyUI-Cloud-Offload Updated
#38
Record M4 fast restore production proof
jethac/cloud-offload Updated
#37
Add background cache scrub and volume recovery
jethac/cloud-offload Updated
#36
Add signed prepared cache trust receipts
jethac/cloud-offload Updated
#35
Record accepted M3 lease closure evidence
jethac/cloud-offload Updated
#34
Add revocable leases and provider billing closure
jethac/cloud-offload Updated
#8
Show lease closure and paid runtime controls
jethac/ComfyUI-Cloud-Offload Updated
#7
Add persistent Cloud Jobs panel
jethac/ComfyUI-Cloud-Offload Updated
#33
Add reloadable Cloud Jobs visibility
jethac/cloud-offload Updated
#32
Use measured history for complete GPU estimates
jethac/cloud-offload Updated
#31
Record accepted M1 cleanup replay
jethac/cloud-offload Updated
#30
Fix paid worker cleanup race
jethac/cloud-offload Updated
#6
Bind rental confirmation requests to ComfyUI API
jethac/ComfyUI-Cloud-Offload Updated
#29
Recognize managed worker authentication in preflight
jethac/cloud-offload Updated
#28
Record merged M1 confirmation UI
jethac/cloud-offload Updated
#5
Add GPU rental confirmation
jethac/ComfyUI-Cloud-Offload Updated
#27
Enforce rental confirmation policy
jethac/cloud-offload Updated
#26
Bind paid partition launch to preflight
jethac/cloud-offload Updated
#25
Add read-only partition preflight
jethac/cloud-offload Updated
#24
Close M0 with redacted production evidence
jethac/cloud-offload Updated
#23
Extend the bounded corruption observation window
jethac/cloud-offload Updated
#22
Fetch assigned manifests through the coordinator
jethac/cloud-offload Updated
#21
Make corruption canaries unique per campaign
jethac/cloud-offload Updated
#20
Write corruption canary bytes only once
jethac/cloud-offload Updated
#19
Match corruption canary launch profile identity
jethac/cloud-offload Updated
#18
Resolve corruption manifests by exact identity
jethac/cloud-offload Updated
#17
Use a fresh object for corruption canaries
jethac/cloud-offload Updated
#16
Make restart canary prove replay and closure
jethac/cloud-offload Updated
#15
Fix Windows restart canary and consolidate goal evidence
jethac/cloud-offload Updated
#14
Prepare corruption canaries before Pod launch
jethac/cloud-offload Updated
#13
Add reversible production failure canaries
jethac/cloud-offload Updated
#12
Make benchmark cache state authoritative
jethac/cloud-offload Updated
#11
Complete the Cloud Offload goal record
jethac/cloud-offload Updated
#10
Declare the runtime keyring dependency
jethac/cloud-offload Updated
#9
Force fresh execution in production benchmarks
jethac/cloud-offload Updated
#8
Add spend-capped production benchmark harness
jethac/cloud-offload Updated
#7
Make the job journal lifecycle-authoritative
jethac/cloud-offload Updated
#6
Add idempotent JobEventV2 journal transport
jethac/cloud-offload Updated
#5
Document the Cloud Offload product goal
jethac/cloud-offload Updated
#4
Show cloud startup and cache progress
jethac/ComfyUI-Cloud-Offload Updated
#4
Complete durable prepared storage and startup feedback
jethac/cloud-offload Updated
#3
Add prepared storage controls
jethac/ComfyUI-Cloud-Offload Updated
#3
Add storage-aware prepared state
jethac/cloud-offload Updated
#2
Support boxed subgraphs and remote HF models
jethac/ComfyUI-Cloud-Offload Updated
#1
Cancel cloud jobs when execution stops
jethac/ComfyUI-Cloud-Offload Updated
#2
Define storage-aware cloud offload
jethac/cloud-offload Updated
#1
Fix cloud offload worker reliability
jethac/cloud-offload Updated
A note on TRELLIS.2: ComfyUI now supports it in core. The historical work here concerns quantized/GGUF-safe weight handling and projection-attention coverage; it is not a claim to have created core TRELLIS.2 support.
Statuses are a dated snapshot. Each title links to the current pull request on GitHub.