Omnigen is out and it is a AI solution that allows you to edit images and do some advanced Photoshopping or Image manipulation without even understanding what a Layer is. This is a 2 Part tutorial. I will be teaching you how to install and use Pinokio to manage multiple AI tools and install them with one click. This is how we will install Omnigen and use it.
With Omnigen you can ask AI to Deblur, add some Bocca, Soften, Recolor, add texture, combine 2 persons from 2 different images, remove blemishes from a face, make someone thinner, and do anything Photoshop can do. This is next level AI art Manipulation and it's ABSOLUTELY FREE!
Omnigen is created by VectorSpaceLab and you can see their github page here which will have instructions on how to install this manually - VectorSpaceLab/OmniGen: OmniGen: Unified Image Generation. https://arxiv.org/pdf/2409.11340
What can Omnigen do? Statement Directly from Developers
"OmniGen is a unified image generation model that you can use to perform various tasks, including but not limited to text-to-image generation, subject-driven generation, Identity-Preserving Generation, image editing, and image-conditioned generation. OmniGen doesn't need additional plugins or operations, it can automatically identify the features (e.g., required object, human pose, depth mapping) in input images according to the text prompt. We showcase some examples in inference.ipynb. And in inference_demo.ipynb, we show an interesting pipeline to generate and modify an image.
"
From Omnigen's website
In the examples above the Boy at top was changed into a woman reading a book which the very same pose. This eliminates the need for using photoshop, controlnet, and advanced tools. You simple need to describe what you want and give it examples.
In the picture below that, Omnigen was asked to take the man in the middle of the photo and combine him with the boy in the second picture and show them reading a book and it did so perfectly, altering the position of his hands and sitting next to the boy in a library. AMAZING!
For this tutorial we will be using Pinokio to simplify the installation. Pinokio will allow you to install and manage AI tools like Omnigen, StableAudio, ComfyUI, Fooocus, Live Portrait, Hallo and many more tools with the click of a button. It will install all the PreReq's, updates, and the tool with a single click. AND Yes it is amazing.
You will need to go to https://pinokio.computer/ to access the download link for pinokio.
There will be an installation video that can walk you through the install, but you won't really need it. It will show you the process if you need to see it. Essentially your unzipping the file you download and running the .exe
So click on download which will bring you to the install page.
In the install page you get to choose what type of OS your using. When you click on windows it will not move because it brings you to the part of the installation for windows but you already there. So click on download for windows in step 1 instead.
Now unzip Pinokio and open the folder and there will be a Piniokio Exe file
Just click on it to install
IMPORTANT. YOU WILL get a notice that windows protected your PC. This is common for most tools on Github as they have never been on Windows Store or vetteed through windows. Pinokio has been around for over a year and has proven to be trustworthy.
If you choose to install, Click on "MORE INFO" then click on "Install Anyways". This is actually really common and A111 and ComfyUI required the same thing to install as well as most, if not all, the tools on Github.
Now just wait for everything to install and click ok and next to everything and it will bring up the Pinokio Console. If not, search for Pinokio after it installs and Run it.
Now click on discover and be ready to be amazed!
Basically every hot AI tool on Github can be installed with one click using this tool. It will install the PreReq's, Dependencies, the correct versions of programs, and even make your virtual enviorment and manage it all within the Pinokio/API folder. So as easy as it is to install it will be that easy to manage and uninstall.
Invoke, comfyui, forge, flux, Moshi, Hallo and Live portrait to animate 2d images, Stable Audio and RC stable audio tools to create songs & sound effects, voice cloners like bark, image to video, infill editing tools, ipadapters like instantstyle, even chat bots and even 2d to 3d modeling tools like TripoSR. All installed in a single click.
BUT we came here for Omnigen so find Omnigen which should be at the top right now. You can use the search bar to find it.
Click on it and it will bring you to the download page.
Once you clicked on download you will get this notice
This will create a folder in Pinokio/API named Omigen so click on download on the Save As tab
Now for the Magic. Click on install and Pinokio will do the rest and it's doing a hell of a lot.
This install took me 15-20 minutes. The model that it downloads to generate the art is 15gigs so this might take a while.
BUT once it is done installing click on the popout button at the top.
Note: When I first installed Omnigen I was met with the error:
ENOENT: no such file or directory, stat 'E:\pinokio\api\omnigen.git\{{input.event[0]}}'
Stop the Terminal
Then close out of Pinokio completely and run it again. Then Click on popout once it's fully loaded.
I had trouble with my initial installation of Pinokio and installed it to the default location on Drive C: instead of my Drive E: then I had to restart Pinokio Completely and it came up. BUT i still got an error on generation.
IMPORTANT: The reason I was getting an error in the output image is because the prompt has to be structured in a specific way.
The directions are right above the prompt and you cannot expect the AI to fill in missing information.
You must write it like this: Enter your prompt, use <img><|image_i|></img> to represent i-th input image
Notice that it has image_i , the "i" will have to be replaced with image_1, image_2, or image_3. Once you upload an image it will label the image, which you will see above your image. If you copy the example you will just have to change the "i" to the number of the image. The image below is #1
For example: Girl in <img><|image_1|></img> is riding unicorn in <img><|image_2|></img>
In it's current state Omnigen can take much longer than typical AI art to generators to process your image. There are several factors involved such as the complexity of the image, the original size of the image, the quality your requested.
My first generation combining 2 photos took over 10 minutes which is a very long time. I've waited anywhere from 3 minutes to 20 minutes for a photo so be patient. Keep it simple at first and then try different things. The speed of this tool with improve overtime so make sure to click the update button before you run it each time.
Last tip is to combine photos with like proportion or it will choose one style that might not work well
Comments