From Stills to Motion - AI Image Interpolation in ComfyUI!

what settings would you use if you wanted to keep the output as close as possible to the output image? I've tried to play with the batch creative interpolation across many settings, but no matter what, the image meaningfully changes - just curious if you've been able to accomplish this?

Every time I try comfyui I get a dozen errors that need to be solved.

Always love your vids, awesome! Seeing this error which is similar to other comments, but a little diff, everything's been updated,

Error occurred when executing BatchCreativeInterpolation:


'NoneType' object has no attribute 'lower'

BatchCreativeInterpolationNode.combined_function() got an unexpected keyword argument 'cn_start_at'

Getting this error.. can someone help me please

I keep getting this error when it runs through the batch interpolation node "BatchCreativeInterpolationNode.combined_function() got an unexpected keyword argument 'cn_start_at'" Anyone have any thoughts??

Which one of your workflows is the one you presented here, you have a million different .png files...

seems like most of the plugins in the shared workflow are missing or broken and they are not coming up when searching for custom notes. When loading the graph, the following node types were not found:
Note Plus (mtb)
IPAdapterModelLoader
ACN_SparseCtrlRGBPreprocessor
ACN_SparseCtrlLoaderAdvanced
VHS_LoadImagesPath
ADE_AnimateDiffUniformContextOptions
ADE_EmptyLatentImageLarge
ACN_AdvancedControlNetApply
VHS_SplitImages
STMFNet VFI
KSampler Adv. (Efficient)
ADE_AnimateDiffLoaderWithContext
VHS_VideoCombine
BatchCreativeInterpolation

hi i have a problem: Error occurred when executing KSampler Adv. (Efficient):
'NoneType' object has no attribute 'to'

I'm getting an error regarding the IP adapterModelLoader: 'NoneType' object has no attribute 'lower' :(

Hey, great video :)
I'm trying to find the workflow, I've followed the link, I see loads of workflows but not this one, do you have i direct link to it please? :) Thank you!

Hi there, Thanks a lot for your great tutos! I think there must have been an update that causes the following error ("Error occurred when executing BatchCreativeInterpolation:
BatchCreativeInterpolationNode.combined_function() got an unexpected keyword argument 'positive'). The "Batch Creative Interpolation" node seems a bit messed up after the update.

where do i place the controlnet model in the batch creative interpolater?

Very interesting. Thank you for sharing workflow and tutorial. Unfortunately something is strange: my "Batch Creative Interpolation" node looks different, than the one in the tutorial - there is no "control_net_name" entry to select safetensors file. I tried with loaders but it seems Batch Creative Interpolation does not have an input for that. On SteerableMotion github I noticed there is no input for control net in the INPUT_TYPES - did you do your own modification of SteerableMotion? Or is there sometinh I need to do to make it visible?

POM updated this workflow and the node, which kinda breaks this way of doing it. The new version uses sparsectrl rgb. Def worth a look :) And as always, thank you for all that you do! This workflow helped me out a ton.

Following along with this, where does the stmfnet.pth go? That's a new one for me, and so far, google is useless

Mine is stalling at the box right before output with STMFNet VFI - what is this? I can't find a reference to it in the manager. Thank you!

amazing workflow! which clip vision model get used here? also get an error that it failed to import CuPy any hints? or is it because of my 8 gb gpu?

Thanks for the video! There is one problem. Gives an error message. Gives an error message. How to fix it properly? Also, instead of previewing the image, a black screen is constantly displayed. What to do with it?

""Error occurred when executing BatchCreativeInterpolation:

Error(s) in loading state_dict for ImageProjModelImport:
size mismatch for proj.weight: copying a param with shape torch.Size([3072, 1024]) from checkpoint, the shape in current model is torch.Size([3072, 1280]).""

I updated everything, so the problem is not in the old version.

Very interesting, but I can not find the workflow using the link above, any clue?

Looks amazing! Unfortunately I'm getting an error regarding the IP adapterModelLoader: 'NoneType' object has no attribute 'lower' And a bunch of lines in a file called 'execution.py' that are faulty. Any idea? I updated to the latest version

i want workflow link please , because i am still new in comfyui

This is amazing! I'm using images with very little variation for creating consistent animations. I'm loving this workflow. Thanks for the tutorial!

This is a great tutorial, thank you :) but my question is, is it possible to have this video like a loop? I mean getting a smooth transition from the last image to the first.

Great video mate! Clear, precise and a free workflow too? What's not to like!

Where can i find the workflow

Does this do a good job interpolating images of the same scene/shot? I just want something to animate my images for a movie

Error(s) in loading state_dict for ResamplerImport:
size mismatch for proj_in.weight: copying a param with shape torch.Size([768, 1280]) from checkpoint, the shape in current model is torch.Size([768, 1024]).

no idea what this means or how to fix it?

Great video:) when I loaded the workflow there was no graph image in the preview, and running the prompt it keeps stopping at Batch Creative Interpolation node. any thoughts ? tried loading all missing nodes and restarting,? thanks again and great videos,

can you show how to load this in Comfy UI rather directly showing results

thank you! was waiting for this one. finally got it working but still trying to tame the beast .... not there yet

Question!! My son is doing a project on ''The Mesolithic period'', he wants to use some examples of AI art for his talk on the subject. The problem is, all my attempts using Comfyui are a mutated group of cavemen, he's nervous as it is, he said my AI art will make him the laughing stock. So, are there any simple PROMPTS that i could use to produce good results?. cheers

Hi I truly like your channel and this time I am trying to execute this workflow but I have some issues, pherhaps you may know what is the problem?
I kept having some errors and discovered that changing the ipadapter that you use from ip-adapter-plus-sd15 to ip-adapter-sd15 fixes the issue but still having problems.
After this process the final result is something totally different from the input images, and all the parameters are just like yours in this video.
Do you have any idea what could be the problem I am facing? Because I have no clue at all......I am lost :(

Very cool!! One request, please list the relative paths and exact model names and download links in your workflow page. The developer pretty much says “go and figure it out” on the GitHub page, not much guidance is given. Lots of error messages on first try 😢

you can't load as many images as you like - that's not true. A new model is loaded for each image added and here's the error you get:
WARNING:root:Some parameters are on the meta device device because they were offloaded to the cpu.
loading in lowvram mode 256.0

And nothing happen.
It's a shame, this kind of workflow could have been interesting for long renderings.

thx

Hello my hero. I've been looking for an AI Morph solution for months and never found it, until now. You have already given me a lot of knowledge about automatic1111 and ComfyUI and I am doing a lot of research myself. Yes, I can't see python console anymore but since ComfyUI everything has become so much easier. Merry and relaxing Christmas from the bottom of my heart. AlbertoSono

anyone else getting this error 'ControlNet' object has no attribute 'load_device'?

🔥🔥

Hi! This is two parts question.. Can I use lora in prompt? Like "0" :"<lora:[loraname]:1>". Probably not... So how I can use lora in this workflow? And then, sec question - can I use multiple loras with a connection to timeline "0", "4", "20", "36" etc. in your prompt? Probably not either, then maybe just separate lora for each image? ( &if that possible then probably different model to each other image is possible too? ) Thank you

It looks almost like Deforum extension for A1111 was converted for Comfy 👍

do you need to install cupy seperately?

great.. thank you

keep getting errors whatever i do , what the hell does this means Error occurred when executing ADE_AnimateDiffLoaderWithContext:

module 'comfy.ops' has no attribute 'Linear'......so not worth it

Thanks for the lesson. But I can't get the workflow to work
Error in the sampler, replacing it with a standard one does not help.
Error occurred when executing KSampler Adv. (Efficient):

Expected all tensors to be on the same device, but found at least two devices, cpu and cuda:0! (when checking argument for argument mat1 in method wrapper_CUDA_addmm)

I managed to work through errors on about 5 different nodes (updating comfy in the manager fixed most of them) but am hung up at the end where the animate diff loader feeds into the K-sampler. I have mm_sd_v15_v2.ckpt and sqrt_linear (Animate Diff) set which matches the video. the error is-
"Error occurred when executing ADE_AnimateDiffLoaderWithContext: module 'comfy.ops' has no attribute 'Linear'..."
I wonder if the K sampler it's plugged into has already been updated because I have an additional field called "sampler state" above "add noise" that I don't see in this video.

bro… those girls in TANK TOPS with RAT HEADS are freaking HORRIFYING… 😂

I love this channel so much, it's my go to for latest info on AI!

Out of all your workflows which would be the best first one for a beginner in comfyUI to get started with using? I’m also using an amd 7900xt.

Seems great, but i really struggle to make anything out of it. Tried the settings from video as well as many others and im getting some abominations, flashing images and nothing like source pictures. Any universal settings, that you could recommend?

How did you get your comfy ui to look all colourful?