Commit Graph

  • 3eaa736916 Merge pull request #10 from iGavroche/claude/project-thread-agfj22 main Gavroche 2026-10-20 09:46:40 -04:00
  • a708e844dd v2.3.29: Seed-VC only bridges tonal f0 gaps, not consonants Claude 2026-10-19 20:03:35 +00:00
  • f1a98ee8d1 Strip downscale_ratio tags from KSampler output, matching stock common_ksampler Nynxz 2026-10-16 11:52:32 -04:00
  • 4ff75c45dc Move audio node under ROCm Ninodes/Audio category iGavroche 2026-10-15 11:34:09 -04:00
  • 76b0cde399 Hide legacy node aliases from search to fix duplicate ROCm VAE Decode iGavroche 2026-10-15 11:07:14 -04:00
  • aacaf34562 v2.3.25: Veda R2VA predictor support + trained-budget defaults iGavroche 2026-10-15 10:05:14 -04:00
  • d5e2947946 v2.3.24: Seed-VC air_transfer COLA-correct STFT window iGavroche 2026-10-14 15:10:42 -04:00
  • 57f3908867 v2.3.23: Seed-VC dynamics and air transfer post-process iGavroche 2026-10-14 13:57:07 -04:00
  • 740b347e92 v2.3.22: Sampler attention_backend selector, dead inputs + VAE dtype machinery removed iGavroche 2026-10-14 10:16:55 -04:00
  • c462d83132 v2.3.21: Fix GGML tensor torch_function dispatch on RMSNorm (issue #8) iGavroche 2026-10-13 13:42:21 -04:00
  • 0bac896c02 v2.3.20: Seed-VC vibrato transfer ramps in after note attacks iGavroche 2026-10-13 11:21:12 -04:00
  • a57c96bb4c v2.3.19: Seed-VC fast note runs step instead of gliding iGavroche 2026-10-13 11:03:14 -04:00
  • 2870970926 v2.3.18: Seed-VC singing checkpoint ft_ema_v2, vibrato_transfer, prompt_length, gap bridging, sample-rate fix iGavroche 2026-10-13 10:15:43 -04:00
  • 76e8803717 v2.3.16: Seed-VC progress bar, silent accompaniment, quiet checkpoint loader iGavroche 2026-10-12 14:16:21 -04:00
  • ec4482504d v2.3.16: Seed-VC separate_source + accompaniment output, fix chunk-loop hang iGavroche 2026-10-12 13:39:48 -04:00
  • 2df094dca1 v2.3.15: ROCmSeedVoiceConversion node + per-node docs + veda no-host-sync iGavroche 2026-10-12 08:15:30 -04:00
  • 3ce9da9ae8 Revert h3_tile_size: 512px tiles cause visible seams and over-exposure iGavroche 2026-10-10 16:52:07 -04:00
  • b1fdd1eb52 Add h3_tile_size to ROCm VAE Decode for 1.56x H3 decode speedup iGavroche 2026-10-10 15:34:23 -04:00
  • 1cc9705ad4 v2.3.12: quantize warps tuning + predictor download link iGavroche 2026-10-10 09:33:29 -04:00
  • 0fb43e1615 Use 2 warps for the INT8 quantize kernel on the pointer path iGavroche 2026-10-09 20:26:58 -04:00
  • b8ecf83bdb README: VedaSparseAttention docs and MiniMax H3 benchmark (1.89x sampling on gfx1151) iGavroche 2026-10-09 16:49:09 -04:00
  • ad513845ba v2.3.11: gfx1151 attention launch config + remove env reads flagged by registry scanner iGavroche 2026-10-09 14:56:18 -04:00
  • 522f8366af v2.3.10: add .comfyignore so the registry zip ships only the node payload iGavroche 2026-10-09 11:37:21 -04:00
  • 9e6851e654 v2.3.9: republish after registry flag (reword scanner-trigger comments) iGavroche 2026-10-09 09:50:35 -04:00
  • 4e90df683e v2.3.8: move VedaSparseAttention to ROCm Ninodes/Optimization category iGavroche 2026-10-08 16:37:20 -04:00
  • ee43493d03 v2.3.7: add VedaSparseAttention node (MiniMax H3 block-sparse attention, Triton INT8 on gfx1151) iGavroche 2026-10-08 15:57:07 -04:00
  • 823dcefc68 v2.3.6: registry republish (no functional changes) iGavroche 2026-09-07 11:36:46 -04:00
  • c931e01d3b v2.3.5: fix H3 invisible on older ComfyUI + registry includes iGavroche 2026-09-07 11:35:08 -04:00
  • ad38dd4b99 v2.3.4: add ROCmH3EasyCache for 4-step MiniMax H3 iGavroche 2026-09-03 15:25:04 -04:00
  • fcdd56013c v2.3.3: fix fuzzy tile transitions in ROCmVAEDecodeTiled on causal video VAEs iGavroche 2026-08-27 15:22:41 -04:00
  • 9ee056db51 v2.3.2: fix NestedTensor VAE crash, GGUF K-quants, ComfyUI version constraint, and GitHub issue fixes iGavroche 2026-08-27 14:07:23 -04:00
  • 56aa176594 v2.3.1: fix NestedTensor crash on LTXAV + video VAE memory estimate iGavroche 2026-08-02 17:37:37 -04:00
  • 386a254af5 v2.3.0: ROCm 7.14 support + HIP OOM fixes + naming cleanup iGavroche 2026-07-20 12:35:22 -04:00
  • 8be224cf68 v2.2.9: fix temporal VAE blend seam + extra frames (1393 vs 1201 on 50s LTX) iGavroche 2026-06-28 15:01:24 -04:00
  • 20fae9c790 v2.2.8: re-enable LTX chunked IO (Windows ZLUDA crash fix), 3-way temporal tiling, visible previewer errors iGavroche 2026-06-22 11:16:46 -04:00
  • 8077bd931f v2.2.7: fix Windows access violation in temporal VAE tiling iGavroche 2026-06-22 10:20:43 -04:00
  • 8f4344354f v2.2.6: fix 2x ROCm sampler slowdown on LTX 2.3 video iGavroche 2026-06-15 14:57:09 -04:00
  • e86ff0d518 v2.2.5: Fix LTX-2.3-22b illegal memory access on RDNA 3.5 iGavroche 2026-06-14 12:18:02 -04:00
  • 906cfffd79 v2.2.4: cancel responsiveness for sampler + VAE decode iGavroche 2026-06-14 09:55:33 -04:00
  • 209642b3ff v2.2.3: fix GPU crash on large models iGavroche 2026-06-13 17:48:54 -04:00
  • f56b1fb503 fix: remove all global GPU state changes from ROCMSamplerCustomAdvanced iGavroche 2026-06-13 17:48:24 -04:00
  • f7dc73f5f7 v2.2.2: fix first-frame oversaturation (causal VAE precision), fix vertical banding (fp16 accumulation) iGavroche 2026-06-13 16:36:56 -04:00
  • 298b646ae8 fix: keep causal VAE weights native to prevent first-frame artifacts iGavroche 2026-06-13 15:23:34 -04:00
  • a8ed294a8e fix: remove allow_fp16_accumulation from ROCm SamplerCustomAdvanced iGavroche 2026-06-13 15:20:27 -04:00
  • a78f92f0ca v2.2.1: bump version for publish (2.2.0 already on registry) iGavroche 2026-06-13 11:32:04 -04:00
  • f779d67adf v2.2.0: clean up ROCM SamplerCustomAdvanced interface iGavroche 2026-06-13 11:10:56 -04:00
  • b3b82e1cb8 perf: fix ROCm SamplerCustomAdvanced callback overhead iGavroche 2026-06-13 11:05:27 -04:00
  • d9f8cdbe55 chore: bump pyproject.toml to v2.2.0 iGavroche 2026-06-13 09:56:57 -04:00
  • b4dd9ecc68 chore: bump comfyui_manager.json to v2.2.0 and add new nodes iGavroche 2026-06-13 09:53:37 -04:00
  • 20626b4fd4 v2.2.0: ROCm SamplerCustomAdvanced drop-in with benchmark iGavroche 2026-06-13 09:52:12 -04:00
  • 281472e0e4 v2.1.1: shared arch detection, sampler auto-detect for z-image/ernie/ideogram4 iGavroche 2026-06-12 17:49:56 -04:00
  • c989a48aab v2.1.0: LTX/z-image VAE support, temporal tiling, fp16 default, APU tuning iGavroche 2026-06-12 15:54:56 -04:00
  • 82540c5245 chore(release): bump version to 2.0.11 and update documentation iGavroche 2026-03-08 11:35:08 -04:00
  • ed83aa2f17 docs: Update README for v2.0.10 with new features and improvements iGavroche 2026-03-08 09:44:12 -04:00
  • 64e5ecf00f chore(release): bump version to 2.0.10 for ROCm Ninodes iGavroche 2026-03-08 09:42:28 -04:00
  • a098c72d77 feat: Enhance ROCm Checkpoint Loader with caching mechanism iGavroche 2026-03-05 14:51:48 -05:00
  • 6f88263b13 chore(release): bump version to 2.0.9 and update GGUF loader with architecture detection iGavroche 2026-01-14 11:07:14 -05:00
  • d7a43355d3 fix: LoRA loader API compatibility and remove Unicode emojis (v2.0.9) iGavroche 2025-12-12 16:38:11 -05:00
  • 32ef99d0b2 perf: Add ROCm optimizations and simplify GGUF loader UI (v2.0.8) iGavroche 2025-12-12 16:31:55 -05:00
  • e74fed150e feat: Implement complete GGUF dequantization support (v2.0.7) iGavroche 2025-12-12 16:21:33 -05:00
  • 8c8d915cff fix: Diffusion loader file listing, PyTorch 2.0+ GGUF support, and remove Unicode emojis iGavroche 2025-12-12 12:49:50 -05:00
  • caf33ea526 fix: Windows console encoding error in install.py iGavroche 2025-12-12 12:33:40 -05:00
  • e9d56fb2a0 fix: Add support for multiple model file formats in ROCm Diffusion Loader iGavroche 2025-12-12 12:13:56 -05:00
  • f6c3d39d9f docs: Enhance README with detailed node usage guide for ROCm Ninodes, including VAE Decode, KSampler, and performance monitoring nodes iGavroche 2025-11-30 12:33:04 -05:00
  • 2b7c9e26ee chore(release): bump version to 2.0.3 and update changelog for enhanced KSampler progress reporting iGavroche 2025-11-30 09:10:58 -05:00
  • 8070724ffb chore(release): bump version to 2.0.2 and update changelog for video processing improvements iGavroche 2025-11-27 14:22:26 -05:00
  • ab9ea59779 Fix WAN VAE jitter issue - disable chunking for WAN models (v2.0.1) Nino 2025-11-01 15:02:09 -04:00
  • 5d515f035e docs: add Upgrade to v2 section; fix ROCm branding in title\nchore: add tools/upgrade_to_v2.py migration helper\nrefactor: finalize v2 structure and cleanup wording v2.0.0 Nino 2025-10-30 14:47:52 -04:00
  • db8f9da2c8 chore(release): v1.0.30 - version bump and changelog v1.0.30 Nino 2025-10-29 14:04:05 -04:00
  • 50e84d712b v1.0.29: Fixed video chunk boundary artifacts - Added temporal overlap to prevent darker frames at chunk boundaries Nino 2025-10-28 18:19:51 -04:00
  • 93e2213ec2 v1.0.27: Mature ROCm driver support with gentle memory management Nino 2025-10-19 19:53:41 -04:00
  • 3291a4b00f Add ROCMLoRALoader node with aggressive memory management iGavroche 2025-10-19 18:08:28 -04:00
  • c7b9013eb0 v1.0.22: Introduce ROCMMemorySafeKSampler and enhance memory management iGavroche 2025-10-19 10:35:09 -04:00
  • 7d6d9d2a17 v1.0.21: Fix critical memory calculation errors iGavroche 2025-10-18 18:20:22 -04:00
  • 7e2dd59549 v1.0.20: Optimize ROCM KSampler Advanced for video workflows iGavroche 2025-10-18 18:17:20 -04:00
  • e124941611 v1.0.19: Repository cleanup and version iteration iGavroche 2025-10-18 17:52:59 -04:00
  • 618b715c21 v1.0.18: Correct version numbering from v1.0.17 base iGavroche 2025-10-18 17:43:43 -04:00
  • cf2b7c58ed v1.0.16: Fix critical VAE decode tensor dimension mismatch iGavroche 2025-10-18 17:41:08 -04:00
  • 11dee871cd Iterate version to 1.0.17 Gavroche 2025-10-17 09:11:36 -04:00
  • dedae3d532 Add additional optimizations for Windows hanging issue Gavroche 2025-10-16 19:29:22 -04:00
  • c08ebe0ffa Add safe progress indicators and optimizations for Windows users Gavroche 2025-10-16 19:06:33 -04:00
  • 8e54d04899 Clean up debugging code - WAN VAE frame duplication issue fixed Gavroche 2025-10-16 18:55:14 -04:00
  • 893426f0c6 Fix WAN VAE frame duplication issue Gavroche 2025-10-16 18:49:51 -04:00
  • eb8a3dd5b2 Enhanced debugging for frame count analysis Gavroche 2025-10-16 18:49:38 -04:00
  • 0fc5555adc Add debugging for dark frame issue Gavroche 2025-10-16 18:39:02 -04:00
  • e690f28613 CRITICAL FIX: Remove dangerous memory operations causing Gnome crashes Gavroche 2025-10-16 18:29:27 -04:00
  • edc21f4246 Revert to simpler video processing from v1.0.14 Gavroche 2025-10-16 18:19:53 -04:00
  • 2c5a3dc689 Fix type mismatch in ROCMOptimizedVAEDecode - v1.0.16 Nino 2025-10-12 10:56:35 -04:00
  • 3b90a34214 Iterate version to 1.0.15 Gavroche 2025-10-11 15:56:07 -04:00
  • cae2c60ec8 Fix WAN video decoder darker frame issue Gavroche 2025-10-11 15:53:46 -04:00
  • 0e78fd05f2 Fix WAN VAE chunking logic for frame count mismatch Gavroche 2025-10-11 15:42:19 -04:00
  • 64ae39ddfa Fix tensor value preservation in WAN video decoder Gavroche 2025-10-11 15:38:35 -04:00
  • ba3943ed6e Clean up legacy files and test WAN video decoder fix Gavroche 2025-10-11 15:24:31 -04:00
  • 150166c206 Fix WAN video decoder batch processing issue Gavroche 2025-10-11 15:22:25 -04:00
  • 528cc8423f v1.0.14: Advanced KSampler optimizations for gfx1151 Gavroche 2025-10-11 14:44:14 -04:00
  • 12f987c57b Fix SaveImage data type error and CLIP validation for Windows compatibility Gavroche 2025-10-11 14:22:42 -04:00
  • 29b6eb293d Add ComfyUI management scripts and process control workflow Gavroche 2025-10-11 12:00:28 -04:00
  • 2bf800f763 🔧 FIX DEBUG SYSTEM: Handle torch import gracefully Gavroche 2025-10-11 10:42:46 -04:00
  • 4d70c04bd6 🔧 INTEGRATE DEBUG SYSTEM: Add comprehensive data capture to ROCM nodes Gavroche 2025-10-11 10:36:24 -04:00
  • b3f0d7bd8d 🚨 CRITICAL FIX: Remove debug code causing 100s VAE decode regression Gavroche 2025-10-11 10:21:30 -04:00