This workflow allows you to morph text with different background images.
Based on Latent Vision's Youtube video here:
Models needed
1) Your favourite SD1.5 Checkpoint. Put in the models/checkpoints folder
2) Controlnet. control_v1p_sd15_qrcode_monster_v2.safetensors. Download from here: https://huggingface.co/monster-labs/control_v1p_sd15_qrcode_monster/blob/main/v2/control_v1p_sd15_qrcode_monster_v2.safetensors and put in models/controlnet
3) AnimateLCM. https://huggingface.co/wangfuyun/AnimateLCM Download AnimateLCM_sd15_t2v_lora.safetensors and place in the models/loras folder. Download AnimateLCM_sd15_t2v.ckpt and place in the models/animatediff_models folder
4) IPAdapter Models. I am asking you to install slightly more than what is needed so that if you change some of the options, it will still work. If you are getting errors, you will have to install the exact model that you are missing.
CLIP-ViT-H-14-laion2B-s32B-b79K.safetensors and CLIP-ViT-bigG-14-laion2B-39B-b160k.safetensors. Install using the manager, search for "clip".
ip-adapter_sd15.safetensors, ip-adapter_sd15_light.safetensors, ip-adapter_sd15_vit-G.safetensors, ip-adapter-plus_sd15.safetensors. Install using the manager, search for "ipadapter"
Instructions on Workflow
1) Update your comfyui and install manager (https://github.com/ltdrdata/ComfyUI-Manager)
2) Load the workflow and install missing nodes
3) Ensure the correct models are loaded in the teal nodes
4) Tune settings in the blue nodes to your liking. Explanation in workflow.
Description
FAQ
Comments (18)
You are really good👍
Credit goes to matt3o, the author of the IPAdapter ComfyUI integration. He puts out great work for the nodes and tutorials on YouTube under the Latent Vision channel. Please subscribe to his YouTube channel and star his GitHub repository.
@PixelMuseAI I will. Have a nice day👍
Thanks. What 1.5 model did u use? My results are not so good as yours
https://civitai.com/models/132632/epicphotogasm
This was the model that I used.
If you need further help, you can upload the input files you used to Google drive, send me the details on models and settings and I can help to debug.
been using this for awhile, but recently has not been working as it has previously. maybe a comfyui update changed something. the text seems to disappear now and then reappear. Could you take a peek? maybe a setting that i am missing
users are reporting that some functionality in this workflow is broken recently. i will need some time to debug.
Love this. Thanks! Stupid question, sometimes the text does not appear at all- any ideas? Also, in the prompt, is this a prompt describing the background images or the text? Thanks!
users are reporting that some functionality in this workflow is broken recently. i will need some time to debug.
The prompt is to describe the background image. The text is controlled by the input images.
@PixelMuseAI Awesome, thanks. I found that changing the IPadapter weight from 1.0 to 1.1 often resolves the issue
@hahsdakiiasda thanks for your help. I'll do some testing and will update the workflow with the higher Ipadapter weights as default soon.
@PixelMuseAI Awesome, thanks. It seems to really be random though- perhaps seed dependent. Sometimes the text shows up, sometimes not. Trying to find a variable that actually impacts the final, but I'm in way over my head at this point. Thanks for your work/help!
Hi, when using two pictures it works as in the example, but if there are all five in the folder, everything blurs and the text is distinguishable through the picture. Can you tell me what the problem is?
The technique should work with more than 2 photos, as per this video here: https://civitai.com/images/11256261
Can you upload the photos you are using to a file sharing service so I can help you to debug?
How can I configure this such that it just produces a single effect for the entire duration instead of cycling through them?
Another question is how can I similarly to above, just use a single image file as a mask?
There are 2 x load images (path) nodes. I have not tried this myself, but you can try adding 2 copies of the same image in both the controlnet and images folder.
I've tried that, it still fades out and back in to the individual images like 5 of the same scene. I'm trying to remove the transition
Ciao Matteo, the workflow crashes every time i go to the left side of it, i tried diffrent comfy ui versions, do you know what it might be?