v0.1.4

Breaking Changes

• All booleans were using a custom widget (BOOL) we now switched to the BOOLEAN one since now there is builtin support for convertible widgets (comfyanonymous/ComfyUI@9534f0f). All the examples were updated accordingly.

• Remove the Prores node and replace it with a more general ffmpeg streaming node. (still not fully tested outside it's old prores mode)

What's Changed

• Use PIL for GIF and not ffmpeg (better quality / filesize overall)
• Add interpolation options to the Fit Number node to interpolate the value ( maths from easings.net):
  
• Add padding support to Transform Image (using F.pad)
  
  

PR merged

• refactor: 💄 august by @melMass in #50

Full Changelog: v0.1.3...v0.1.4

v0.1.3

Hopefully starting with v0.2.0 releases won't be just about install issues.

What's Changed

• fix: 🔥 manage pip from install only, remove requirements.txt by @melMass in #38

Full Changelog: v0.1.2...v0.1.3

v0.1.2

Note

This is mostly about fixing #35

What's Changed

• Update README by @melMass in #33
• Fix unclickable gallery buttons on firefox by @dougwhite in #34
• fix: 📝 last release by @melMass in #36

New Contributors

• @dougwhite made their first contribution in #34

Full Changelog: v0.1.1...v0.1.2

v0.1.1

See the previous release note

What's Changed

• dev/next by @melMass in #32

Full Changelog: v0.1.0...v0.1.1

v0.1.0

Warning

this is a breaking change release

New Nodes

• Save to Gif: Basic node to save the images from the batch into a GIF, with preview over the node:
  

New types & widgets

• Color is now working properly:

• A "real" BOOL type with an associated widget (might switch to using regular litegraph toggles in the future) :

What's Changed

• Restore Face node by @melMass in #8
• feat: ✨ FILM interpolation nodes by @melMass in #11
• docs: 📝 update readme by @melMass in #15
• chore: ✨ add config for issue templates by @melMass in #25
• feat: 🚀 release 0.1.0 by @melMass in #22
• Update requirements-wheels.txt by @melMass in #30
• fix: 🚑️ TF by default fills vram
• feat: 🎨 add support for image.size(0) == 0
• feat: 🔥 add batch support for load image sequence
• docs: 📝 add jp and cn (using deep translation)
• feat: ✨ extract node names using ast
• feat: ✨ WIP batch from history
• fix: ⚡️ from tensor2np always returning a list
• feat: ✨ added a way to export the node list
• fix: ⚡️ a few missing doc
• feat: 🚀 add a few examples
• docs: 📝 add preview for examples
• docs: 📝 add model notice
• docs: 📝 add more description to examples
• docs: 📝 fold each comfy mode
• chore: ✨ before categorize
• feat: ✨ categorize
• feat: 🔥 add API endpoints
• feat: ✨ add FitNumber node
• fix: 🐛 tensor2pil
• fix: 🐛 crop nodes
• feat: 🚀 debug node
• refactor: 🚧 remove color_widget
• feat: ✨comfy_widget shared utils
• fix: 🔥 deprecate some nodes and fix image list
• fix: ✨ Various widgets issues
• feat: 🚧 add my CLIs
• refactor: ♻️ remove test
• fix: 🎨 refactor and add Gif preview on node
• fix: 🐛 add base64 prefix to outputs
• refactor: ♻️ removes a few nodes, moved other around
• fix: 🚑️ set debug level from endpoint
• fix: ⚡️ update node_list
• fix: 🎉 remove tests & add missing docs
• feat: 🚀 add install script
• fix: 🐛 install logic
• fix: 🚑️ always use latest for now
• fix: 💄 save gif issues
• test: 🔧 pipe detection
• fix: 🚀 try to support remote install too
• fix: ✨ install process tested in comfy-manager (embed, colab)
• fix: ✨ also try to copy web if symlink fails
• feat: ✨ update install instructions

Full Changelog: v0.0.1...v0.1.0

v0.0.1

Initial release

The deps-wheels archives are prebuilt wheels of the following packages, meant for the python-embed mode of ComfyUI. I don't use it myself but I tested the install process and nodes inside it!

• insightface
• mmcv
• easydict
• pycocotools

Don't hesitate to report issues

What's Changed

• ci: 🚀 add CI scripts (testing) by @melMass in #4

New Contributors

• @melMass made their first contribution in #4

Full Changelog: https://github.com/melMass/comfy_mtb/commits/v0.0.1