H4 Image2Video
***Update***
This workflow has been throwing exceptions, as well as for some reason it'll save
each individual frame opposed to combining them into a singular gif.
This is something on tensors end, the workflow works phenomenally locally.
Which means that this can occasionally generate all the frames OR throw
an exception. - Updated the Workflow, made some changes that were causing excessive
duplication of images to be generated without combining.
They still do this on occasion, I dunno why, hopefully Tensor can sort this out soon.
Also, for some reason the option for pingpong isn't showing up in the tool.
As such I've left pingpong on, so it'll play forward then in reverse
***Update***
This ones a bit of a complex workflow, it has 4 individual sampling steps
in order to apply detail step by step. It uses animateddiff to apply the animation
steps to the original image. The end result is an animation effect.
Sometimes it comes out epic just as you expected, other times it just applies
animated effects. This is because the model you choose can't interpret the original
image during the img2img phase of this workflow.
That doesn't mean it wont work, it just means it may not turn out how you expect.
Keep an open mind and just have fun with this. On that note:
What else does this mean? It means that this can throw a lot of exceptions right now,
it's not the workflow, it's Tensor. (No disrespect)
The concept behind this is simple
Choose an image you want to animate or apply animations too(is what I should say).
Enter a prompt, try to stick to the image as much as possible, it is possible for this to generate
a completely random image if the prompt doesn't match. Then generate.
Few things;
This workflow can take some real time right now on tensor.
This is a hardware restriction not a workflow one, that being said be patient.
It's possible for this to throw exceptions again I say , it's not the workflow.