I wanted to give it a proper try before replying. This is what I was looking for. Detailed searches, custom tags (I can use them as various quality ratings) and more. It’s also nice to know it can run as a standalone executable in case I want to try out AuraFlow before it becomes available in A1111/Forge. Thank you!
The schnell live demo is impressive. My decade-old rig takes minutes per step on Flux, so this is actually a useful way for me to check for known concepts in almost real-time. It even had time to show me a quick Phở when I was typing in photo.
I didn’t even think of dual cards, because I have an old & budget motherboard with one slot. But 2 x 16GB GPUs and a new motherboard (and if necessary, new CPU) and PSU and it might even still be cheaper than a 24GB NVIDIA for me. Of course I’d have to explore the trade-offs in detail because I’ve never looked into how dual cards work.
(but truth be told, I just as easily could settle for a 1x16 GB if I’m confident it would be able to train, even if slowly, AuraFlow or FLEX LoRas for the upcoming Pony v7 model. It’s just a hobby.)
I am using a lot of Pony* models at the moment and Pony v7 looks like it will switch to AuraFlow or FLUX [1] so it’s useful to hear your experience with it on a 3080Ti.
Good to know about CUDA/Direct ML.
I found a couple of 2022 posts recommending 3090s, especially since cryptocoin miners were selling lots of them cheap. Thanks for the heads up about the 5000 release, I suspect it will be above my budget but it will net me better deals on a 4090 :P
Do you know how much wattage those GPUs use, even if I disconnected my solar panels and ran the card 100% 24/7? Protip: it rounds down to zero.
If you’re serious about the global environmental crisis, comrade, organize with others to fight industrial-scale culprits instead of wasting your valuable time blaming trivial people.
Changed my mind lol
I haven’t determined that. I only have one device set up to run SD and haven’t organized any test with someone else.
I mean, if the concern is that tiling is a factor relative to non-tiling, okay, but if someone else is tiling, l’d think that they’d get the same output.
That’s true, I’ll check to see if the metadata mentions the tiling was used.
Good call-out. My (naïve) understanding is that tools like tiling VAE to handle low VRAM, and lowing steps in the more stable of the samplers, are going to have a generally negative impact on the result, and a very similar image with better detail could be remade using similar variables on better hardware. Maybe that’s a bit idealistic. Like you said, the seed mode usually changes images with size. (You said ‘usually’, is there a way to minimize this?)
edit: I’m aware ‘better’ and ‘higher quality’ are vague and even subjective terms. But I’m trying to convey something beyond merely higher resolution.
The lack of forced monetization is why I joined Lemmy. It’s a feature.
I just want to say the maintainer who wrote that appears to have handled this gracefully. It gives me hope.
They’ve made a transparent public announcement, making it clear what we should and shouldn’t expect from them, and how we should handle it. They understand the FOSS paradigm (no, I correct myself, the digital paradigm) and have given their blessings for the community to do what they do best. I’d guess the smart thing to do is play along with the cease notice to avoid consequences, go underground and make YouTube play whack-a-mole with sock-puppets and hostile jurisdictions.
Cut off one head and three shall take its place. Wind in your back lads, wherever you go.
Seeing the previous krita-ai-diffusion demo and the skill of the demonstrator in selecting and img2img’ing encouraged me to give Krita a second try (it turns out my laptop+Mint had a keyboard setting which made the panning shortcut break, so I’ve fixed that and now it’s far more usable. It was ‘disable touchpad when typing’, so Space wouldn’t pan with the touchpad).
As for this video, it’s great to see the ComfyUI integration (and with a logo on the node too!). The tool was already powerful and impressive, and this seems like a huge step up.