Is your feature request related to a problem? Please describe.
Recently released Deforum SD lets you create videos from a text prompt, which changes over time given multiple prompts that activate after x amount of frames.
Describe the solution you'd like
The ability to add the required models (AdaBins, dpt_large-mids) to the models/pretrained folder, a new tab in the webgui called "txt2video" that has an option for a starting prompt (including the ability for negative prompt), or a starting img with prompt.
An add/remove button to add additional prompt segments at X frame with settings to change the CFG as well as the denoise, with a slider to switch between prompt inputs, and an additional input field for how many frames/seconds you want till the new prompt is activated over the previous one.
Describe alternatives you've considered
Using the actual Deforum Release. Simple implementation as a script for both txt2img and img2img, however I'm not sure how to do this myself as the deforum stable diffusions claims 2 additional models are needed besides SD.
Is your feature request related to a problem? Please describe. Recently released Deforum SD lets you create videos from a text prompt, which changes over time given multiple prompts that activate after x amount of frames.
Describe the solution you'd like The ability to add the required models (AdaBins, dpt_large-mids) to the models/pretrained folder, a new tab in the webgui called "txt2video" that has an option for a starting prompt (including the ability for negative prompt), or a starting img with prompt.
An add/remove button to add additional prompt segments at X frame with settings to change the CFG as well as the denoise, with a slider to switch between prompt inputs, and an additional input field for how many frames/seconds you want till the new prompt is activated over the previous one.
Describe alternatives you've considered Using the actual Deforum Release. Simple implementation as a script for both txt2img and img2img, however I'm not sure how to do this myself as the deforum stable diffusions claims 2 additional models are needed besides SD.
Additional context https://www.youtube.com/watch?v=rvHgcOa9gDk https://github.com/HelixNGC7293/DeforumStableDiffusionLocal