I wanted to play around with the recently released Omnigen model, so I whipped up a quick Workflow to load 0-3 images (batch or manual) and load them in to the generator. Nothing too fancy, but it does have a text node with instructions on how to get the model, if the autodownload fails.
Description
FAQ
Comments (11)
please can you advise where to get omnigen node? Its not on the comfyui manager.
Hunh it shows up for me: the node name in manager is "OmniGen - ComfyUI" by AIFSH. If that's not showing up, the github is at https://github.com/AIFSH/OmniGen-ComfyUI
does even the directinstall via manager --- github url not work? https://github.com/AIFSH/OmniGen-ComfyUI if so just download it into your costum nodes-folder (pref. with git clone *url*) and launch comfy
Hello, on the OmniGenNode the "latent" node is not connected...
On your workflow the OmniGenNode has no latent input?!?
@nanunana The latent input is there, but nothing is connected to it.
Looks like they made some changes last night (https://github.com/AIFSH/OmniGen-ComfyUI/commits/main/), including removing width/height and making it take a latent input instead. Interestingly, unlike other models, it doesn't seem to accept batches: making the empty latent batch size 3, for example, doesn't make it produce 3 output images. Odd.
I've updated this workflow to V2 with the latent change; give that a shot.
@mwoody450 i will try...
@mwoody450 now i get a error i don´t understand, i will post a picture what i want to do. The error is "total images must be the same as the number of image tags, got 0 image tags and 2 images" i take this image tags: |image_1|,|image_2|
@nanunana Your prompt must contain as many image_# (where # is the number of images) equal to the number of images you've given it, which you control by enabling or disabling the image nodes in the workflow. You do not need the || marks. If you've not enabled any images, image_ should not appear in your prompt.
@mwoody450 but i have 2 images activated and 2 images in the prompt, can't you even post a .png in which 2 or 3 images are included. I could see from the workflow where my error lies.