ComfyUI IPAdapter V2 style transfer workflow automation

Comfy-UI IPAdapter update
@latentvision, huge thanks for matteo - the creator!!!
In this video we will see how you can transfer a style from one image to other images with the help of the ipadapter v2, since we will want the style mainly on a background, we will also work with masks. And in addition we will see how the entire workflow can be automated
#comfyui #stablediffusion #ipadapter #mask #automation
follow me @ / pixeleasel
workflow
drive.google.com/file/d/1JdO4...
juggernaut model
civitai.com/models/133005/jug...
IPAdapter GitHub
github.com/cubiq/ComfyUI_IPAd...

[@AlistairKarim]

Short and on point, real awesome tutorial. Thanks for sharing!

[@PixelEasel]

thanks for commenting!!

[@knightride9635]

Subscribed! Nice workflow, thank you

[@PixelEasel]

thanks!

[@DJVARAO]

Awesome tutorial! Thank you!😁

[@PixelEasel]

thanks! awesome comment 😊

[@35wangfeng]

Awesome work! Thanks for sharing!

[@PixelEasel]

thanks for commenting!

[@ronnykhalil]

thank you! super helpful and just what i was looking for

[@PixelEasel]

🥳 nice to know! thanks for the comment!

[@typogram4297]

Great tutorial. Thank you.

[@PixelEasel]

thx!!!

[@hmmrm]

Thanks for sharing!

[@PixelEasel]

thank for commenting 😊

[@pto2k]

Great workflow idea. Thanks. I had that same error with BatchCLIPSeg others mentioned. I found that it can be replaced with the CLIPSeg Masking (WAS Node Suite) node + ToBinaryMask (ImpactPack) node. That functions the same I think.

[@PixelEasel]

thanks a lot! I'll check it

[@jorgvollmer5971]

The same here, it works well with the Clipseg Masking.

[@3dpixelhouse]

Fantastic workflow! But... where can I find the Batchclip-Node? The Manager dosn't find it.

[@PixelEasel]

thanks! here's the link github.com/kijai/ComfyUI-KJNodes

[@FeyaElena]

Thanks for the informative video! I am getting There was an error while executing BatchCLIPSeg: The input and output should have the same number of spatial dimensions, but the input received a tensor with spatial dimensions [1, 352, 352] and the output dimension is (832, 1216). Please provide the input tensor in the format (N, C, d1, d2, ...,dK) and the output dimension in the format (o1, o2, ...,oK). Can you tell me what I need to fix?

[@PixelEasel]

did you change anything? upload an image?

[@jomiller7332]

Yep I get the same message , must be something with BatchClipseg , because without that Node its fine

[@AlexDisciple]

Another question, what does "Stop_at_clip_layer -1" do?

[@PixelEasel]

it's set the last layer for the diffusion process - in other words, it doesn't skip any layer of the clip. it is a parameter that is provided by the creator of the checkpoint

[@AlexDisciple]

Thanks for this. I get the same error with BatchClipSeg "Input and output must have the same number of spatial dimensions, but got input with spatial dimensions of [352] and output sise [832,1216]. Please provide input tensir in (N, C, d1, d2, ....,dk) format and output size in 9o1,o2, ...,oK) format"

[@PixelEasel]

I received a few more responses from people who were unable to work with this node. I still haven't found an answer to this, unfortunately

[@JoyHub_TV]

In the "IPAdapter Tiled" node I don't have the weight option for "Style Transfer (SDXL)" but i have "style transfer" and "style transfer strong." how do i get the "Style Transfer (SDXL)" option?

[@PixelEasel]

it's probably because you have the latest version which is good. you can try both

[@ZergRadio]

at 1:52 where do I get this 1.5 model XL.safetensor and where do I put it? When I have the Load CLIP Vision up and use the pull down menu there is no model XL

[@PixelEasel]

you need to put the clip vision models in this root ComfyUI\models\clip_vision an here you can find all the needed model

[@ZergRadio]

@@PixelEasel Thanks for the reply. I have 4 clip models in the right folder. But my other question is where do I get the "model" called model.safetensors from? As I do not have a model by that name in the root folder. So where can I download that specific model which you show at 1:52. I see that there are two models names model XLsafetensors and model.safetensors. I do not have them in my root folder.

[@ideasinspiration8231]

Thanks for this amazing tutorial! I am running into error: Following your workflow once i run it I get error on BatchCLIPSeg: Error occurred when executing BatchCLIPSeg: Input and output must have the same number of spatial dimensions, but got input with spatial dimensions of [1, 352, 352] and output size of (832, 1216). Please provide input tensor in (N, C, d1, d2, ...,dK) format and output size in (o1, o2, ...,oK) format. Any idea how to fix?

[@PixelEasel]

thanks! you need to check the size of the image you connect

[@ideasinspiration8231]

@@PixelEasel I am using the workflow I download from you. and the size is the image you have set in the workflow.

[@pedroquintanilla]

@@PixelEasel

[@pedroquintanilla]

El mismo problema

[@PixelEasel]

weird . I'll check if I can figure out why

[@naderreda8523]

unfortunately i got the same error , Please provide input tensor in (N, C, d1, d2, ...,dK) , any updates ?

[@PixelEasel]

which node is turning red?

[@Gifttv204]

Please make video on bg colour matching workflow ❤

[@PixelEasel]

what do you mean? how to match colors for new background?

[@Gifttv204]

@@PixelEasel yes

[@PixelEasel]

I will...

[@pedroquintanilla]

problem Error occurred when executing BatchCLIPSeg: Input and output must have the same number of spatial dimensions, but got input with spatial dimensions of [1, 352, 352] and output size of (832, 1216). Please provide input tensor in (N, C, d1, d2, ...,dK) format and output size in (o1, o2, ...,oK) format. File "F:\ComfyUI_windows_portable_nvidia_cu121_or_cpu\ComfyUI_windows_portable\ComfyUI\execution.py", line 151, in recursive_execute output_data, output_ui = get_output_data(obj, input_data_all) ^^^^^^^^^^^^^^^^^^^^^^^^^^^^^^^^^^^^ File "F:\ComfyUI_windows_portable_nvidia_cu121_or_cpu\ComfyUI_windows_portable\ComfyUI\execution.py", line 81, in get_output_data return_values = map_node_over_list(obj, input_data_all, obj.FUNCTION, allow_interrupt=True) ^^^^^^^^^^^^^^^^^^^^^^^^^^^^^^^^^^^^^^^^^^^^^^^^^^^^^^^^^^^^^^^^^^^^^^^^^^^ File "F:\ComfyUI_windows_portable_nvidia_cu121_or_cpu\ComfyUI_windows_portable\ComfyUI\execution.py", line 74, in map_node_over_list results.append(getattr(obj, func)(**slice_dict(input_data_all, i))) ^^^^^^^^^^^^^^^^^^^^^^^^^^^^^^^^^^^^^^^^^^^^^^^^^^^ File "F:\ComfyUI_windows_portable_nvidia_cu121_or_cpu\ComfyUI_windows_portable\ComfyUI\custom_nodes\ComfyUI-KJNodes odes.py", line 2257, in segment_image resized_tensor = F.interpolate(tensor, size=(height, width), mode='bilinear', align_corners=False) ^^^^^^^^^^^^^^^^^^^^^^^^^^^^^^^^^^^^^^^^^^^^^^^^^^^^^^^^^^^^^^^^^^^^^^^^^^^^^^^^^ File "F:\ComfyUI_windows_portable_nvidia_cu121_or_cpu\ComfyUI_windows_portable\python_embeded\Lib\site-packages\torch n\functional.py", line 3934, in interpolate raise ValueError(

[@PixelEasel]

did you change something?

[@ojciecvaader9279]

I have the same problem. Changed: Im not sure if I have correct clip vision model. I have updated IPadapter lately and those clip vision models became obsolete

[@PixelEasel]

so maybe try to reinstall..