To Do List

[Trentonom0r3]

• Fix IMG2IMG Inpaint

• Add Script Arg controls

• Marker Generation Option

• LORA options for prompts panel

• Temporal consistency features

• Grid Creation, style transfer, etc.

• Seed Finder

• Wiki info

[Trentonom0r3]

• Fix IMG2IMG Inpaint

• Add Script Arg controls

• Marker Generation Option

• LORA options for prompts panel

• Temporal consistency features

• Grid Creation, style transfer, etc.

• Seed Finder

• Wiki info

Img2img inpaint fixed

Script arg controls added

• need to add dynamic visibility & implement values in API call.
• outside of this, pretty much set up to go. Shouldn't be too intensive.

Marker generation option

• in progress, adding option to simply generate from markers, as well as option to create grid from markers.

LORA options

• idk why I haven't done this yet, it's a super simple implementation. Probably on the back burner for now as I don't really use LORAs

Temporal consistency features

• grid creation needs refining, but we're at a good initial state
• implementing "send to ebsynth" button, which will send keyframes and input to ebsynth, you just have to run the synth.
• (later) implementing a custom ebsynth method using the ebsynth GitHub project, and RAFT optical flow. I'll probably post this as a separate repo as it can work standalone as well as with the extension.

Seed finder

• for this, I'm adding batch and size options to the c++ effect.
• by default, generation will still be ran based on layer size & single generations
  • if you enable width/height sliders, the extension will use those instead.
  • if batch size/batch count is > 1, the extension will output generated images to a grid image gallery instead of back to layers. (This is under the assumption that if batch size > 1, you're trying to test and find seeds)

Wiki info

• in progress, need to add more detailed info,

After implementing the rest of these, the extension will pretty much be fully featured, and I can work on refining the workflow, exploring new, more efficient methods, and perhaps creating an extension for webUI that could enable extra features for the AE extension itself.