VQGAN-CLIP VS stable-diffusion

VQGAN-CLIP

I've screwed with many text-to-image models over the past couple of years, and I found that while I currently enjoy Stable Diffusion's coherency, I have a soft spot for the ImageNet model used by default for VQGAN+CLIP. It easily approaches the uncanny valley when generating people or animals, but makes for great abstract backgrounds and wallpapers. I already have nostalgia for generating images with it on my CPU overnight.

For someone only tangentially familiar with this space, how is this different than e.g. https://github.com/nerdyrodent/VQGAN-CLIP which you can also run at home? Is it the quality of the generated images?

Used https://github.com/nerdyrodent/VQGAN-CLIP

I don't have any of the DALL-Es but I do have a couple from github [1], [2] which gave these outputs[3]

[1] https://github.com/nerdyrodent/VQGAN-CLIP

If you want to try your hand at buggering your whole system - try playing with AI image generation as it uses all possible computer assets :D . There is a lot of forms and installations for those but I VQGANs from github the easiest. Problem is that some require familarity with shell, python and in some cases - you need to enable the Linux subsystem in Windows (is it called a subsystem? it is not exactly a VM). This one is the easiest to install out of all I tried. But I liked the results of Pixray most but I wrecked it. I use this one nowadays.

For those that are interested, I used VQGAN-CLIP, specifically this GitHub repository

Which is a 1-click install of Stable Diffusion with an alternative web interface. You can choose a different approach but this one is pretty simple and I am new to this stuff.

an invisible watermarking of the outputs, to help viewers identify the images as machine-generated.

It sounds like you just want to set up Stable Diffusion to run locally. I don't think your computer's specs will be able to do it. You need a graphics card with a decent amount of VRAM. Stable diffusion is in Python as is almost every AI open source project I've seen. If you can get your hands on a system with an Nvidia RTX card with as much VRAM as possible, you're in business. I have an RTX 3060 with 12 gigs of VRAM and I can run stable diffusion and a whole variety of open source LLMs as well as other projects like face swap, Roop, tortoise TTS, sadtalker, etc...

Use specific GPU on multi GPU systems · Issue #87 · CompVis/stable-diffusion · GitHub

I dealt with this in one of my posts. At least SD 1.1 till 1.5 are all trained on a batch size of 2048. The version pretty much everyone uses (1.5) is first pretrained at a resolution of 256x256 for 237K steps on laion2B-en, at the end of those training steps it will have seen roughly 500M images in laion2B-en. After that it is pre-trained for 194K steps on laion-high-resolution at a resolution of 512x512, which is a subset of 170M images from laion5B. Finally it is trained for 1.110K steps on LAION aesthetic v2 5+. This is easily verified by taking a glance at the model card of SD 1.5. Though that one doesn't specify for part of the training exactly which aesthetic set was used for part of the training, for that you have to look at the CompVis github repo. Thus at the end of it all both the most recent images and the majority of images will have come from LAION aesthetic v2 5+ (seeing every image approx 4 times). Realistically a lot of the weights obtained from pretraining on 2B will have been lost, and only provided a good starting point for the weights.

stable diffusion · CompVis/stable-diffusion@2ff270f · GitHub

Here for example --> https://github.com/CompVis/stable-diffusion