>>29310673
>I'm on a 2070s+32gb of sysram. Not in good place to buy better and even if I did, i don't know if I want to just get more vram, more sysram, or go really spendy and get 5090.
I'm in a similar situation. Going to from 32 to 64 would help a bit since I do max out in some situations when I'm running a browser and other basic stuff on the side but I already have 4 sticks so it would mean replacing them all. Feels like a waste without moving to DDR5 at the same time which would mean a motherboard swap and a CPU upgrade at which point I'd also want to start looking at a new GPU and a new PSU. Basically a full system upgrade at that point.

Do let us know if you figure something out for the eyes or the face. I've been annoyed by that as well.

I wonder how much of a difference there is with 512 and 640 gens.

>>29311398
>You should know that walkanon is cruising on 12GB and started genning at 720p lately. I think you could do it on your 8GB.
Even if it's possible I honestly don't want to increase my gen times. 4-8 minutes is slow enough.

>I also run most of my base images through a face detailer workflow.
What does a workflow like that look like?

>I slapped the upscale by model and upscale model loader nodes right into my Wan i2v workflow
Do you slap the high res image straight into the sampler or do you have a node to resize it first?