12月1月2月3月4月5月6月7月8月9月10月11月
周一周三周五
一年内 265876 次贡献
更少更多
- frostbyte_neo 推送了仓库 github-featured/invoke-ai--invokeai 的 temp-v3.0.0post1 分支比较 5 提交 »cba76c5e54 update version to 3.3.0post18dbaeb74db Updated JS files9e17c61ece Update version to 3.3.0aeeb410435 enable the ram cache slider in invokeai-configure2f9f9c5fa7 closes #4768
- frostbyte_neo 推送了仓库 github-featured/invoke-ai--invokeai 的 speed-up-hashing 分支比较 5 提交 »cb99a5f0e4 Sort results of hashes in _hash_dir to always get the same result, remove dependency on os.walk3bb50926af Testing different file chunk sizesbb3f1b9ca6 Add threading to scan dir calls, cap thread pool in hash function to number of files119d26e102 Remove manual memory management in hashlib_hasher in favor of using Python's built-in bufferinge1c16c33a4 Speed up hashing
- frostbyte_neo 推送了仓库 github-featured/invoke-ai--invokeai 的 separate-graph-execution-loop 分支比较 5 提交 »cd3f5f30dc Run ruff format71ee28ac12 Refractor session runner, move profiling back to processor, create abstract class for session runners, create path for passing in custom session runner to default session processor46c904d08a Rename graph processor to session runner to better describe what it's doing, add before/after callbacks for sessions7d5a88b69d Move graph processor into session_processor_defaultafa4df1991 Separate the logic that actually runs a graph in the session_processor into its own class
- frostbyte_neo 推送了仓库 github-featured/invoke-ai--invokeai 的 saas-target 分支比较 4 提交 »649596cec5 fix 1:1 ratio (#8127)45aa84c01a feat: add user_label to FieldIdentifier (#8126)064d5787c9 Flux Kontext UI support (#8111)d81b23adff fix(nodes): ensure each invocation overrides _original_model_fields with own field data
- frostbyte_neo 推送了仓库 github-featured/invoke-ai--invokeai 的 ryan/upscale-node-prototype 分支比较 10 提交 »59284c707e Change tiling strategy to make TiledStableDiffusionRefineInvocation work with more tile shapes and overlaps.911792f258 Expose a few more params from TiledStableDiffusionRefineInvocation.9567c6e196 Add support for LoRA models in TiledStableDiffusionRefineInvocation.6e47bd14af Add naive ControlNet support to TiledStableDiffusionRefineInvocation9ac9b6a014 Fix ControlNetModel type hint import source.
- frostbyte_neo 推送了仓库 github-featured/invoke-ai--invokeai 的 ryan/tiled-upscaling-3 分支比较 4 提交 »3027e5bf9d Add tqdm logging to the tile loop of TiledStableDiffusionRefineInvocation.4e170baa3a Add step_callback function to TiledStableDiffusionRefineInvocation. This enables preview images and cancellation.e5be627e30 Add the TiledStableDiffusionRefineInvocation.5461de41df Expose LatentsToImageInvocation.vae_decode(...) as a staticmethod.
- frostbyte_neo 推送了仓库 github-featured/invoke-ai--invokeai 的 ryan/tiled-upscaling-2 分支比较 3 提交 »8379feeb8a Refactor TiledStableDiffusionRefineInvocation to more closely mirror TiledMultiDiffusionDenoiseLatents. The biggest improvement is in the handling of the ControlNets - global ControlNet info can now be passed in and it is tiled within the node.b74bc77347 Revert "Revert "Expose vae_decode(...) as a staticmethod on LatentsToImageInvocation.""73f762bb4c Revert "Remove TiledStableDiffusionRefineInvocation. It was a proof-of-concept that has been superseded by TiledMultiDiffusionDenoiseLatents."
- frostbyte_neo 推送了仓库 github-featured/invoke-ai--invokeai 的 ryan/tidy-cross-attention 分支a22331fbe9 Delete unused code for attention map saving.
- frostbyte_neo 推送了仓库 github-featured/invoke-ai--invokeai 的 ryan/seamless 分支比较 2 提交 »9b763b9e4c Fix issue with seamless context managers when seamless is not configured.7f3be627c2 Add more seamless configuration options.
- frostbyte_neo 推送了仓库 github-featured/invoke-ai--invokeai 的 ryan/sd35 分支比较 10 提交 »5edee6997e wip9aaecf5b5c Add utils for inferring SD3 params from a state dict and constructing an SD3 model.b4a2244943 Fix Sd3ModelLoaderOutput name.155bf13d2b Tidy imports in other_impls.py9f7b5f7a85 Miscellaneous cleanup of mmditx.py. Mostly typing fixes.