November Update

Hello friends! First of all, thank you so much for your generous support! There are some really exciting things happening in AI right now, so thanks for hanging in there as I figure out the best path forward.

TLDR: My computer can’t handle generating a 90 second animation locally, so The Wrong Tags RC/BE will be delayed a bit while I generate in the cloud.

So the bad news is obviously that the next project is delayed because my PC sucks. The good news is that with your support I can rent a GPU cloud instance to continue with this animation. Also, I want to explain my PC situation in the interest of full transparency. The bottleneck currently is system RAM. I only have 16GB and need to upgrade to 32GB, but my motherboard is old enough that it only is compatible with DDR4 memory. While I could just buy some DDR4 memory, I will obviously want to upgrade my motherboard eventually, at which point I’ll just need 32GB of DDR5 memory. So, my current plan is to try to upgrade my motherboard, memory, and potentially case sometime this year. This will keep this sort of thing from happening again, and also open the door for me to get a second graphics card at some point so I can generate animations and images even faster!

While I’m generating in the cloud this week, I should be able to use my local PC to work on some requests, and test some new tech as well. Here is the list of things I want to test:

·         Test CogVideoX - Frame interpolation is now available for this open source model. This is huge! I’m not sure if it can do nudity which is why I want to test it. If it can’t, it’s only a matter of time before someone finetunes it with porn. The end goal here is to replace AnimateDiff with this model eventually since the quality is significantly better!

·         Test Moshi - This is the new open source video model that looks comparable to Kling/Runway. Again, just want to test it and see if it can do nudity.

·         Optimize Bigasp - A new version of this model was released recently, and it looks like one of the best NSFW models out there. I would like to test it fully, compare it with the previous version, and post my results/recommendations here in case any of you want to use it.

·         Optimize RealvisXL – This is a very aesthetically pleasing model that I have been using to inpaint my SFW Flux generations. I just need to optimize it, and I’ll post my results/recommendations here!

·         Test SwarmUI – SwarmUI is a front-end (similar to ComfyUI) that combines ComfyUI’s nodes with a simpler UI like Automatic1111. I’m just curious as to the capabilities, specifically for inpainting.