what's the difference between the FLUX.1-Depth-dev and black-forest-labs/FLUX.1-Depth-dev-lora?
what's the difference between the FLUX.1-Depth-dev and black-forest-labs/FLUX.1-Depth-dev-lora?
what's the difference between the FLUX.1-Depth-dev and FLUX.1-Depth-dev-lora?
One is a dish and the other is a seasoning
One is a dish and the other is a seasoning
Instructions unclear, dish stuck in PCIe slot. Teach me to Flux without seasoning.
what's the difference between the FLUX.1-Depth-dev and FLUX.1-Depth-dev-lora?
I'm about to test it but I'd be surprised if it weren't the case that: The Lora is designed to do what you expect a controlnet to do, while the full checkpoint is designed to make people give up on local inference and instead pay cloud inference companies a fortune - but not much more than the Lora.
Lora is there so you don’t need to download a whole new flux model again, you just need to download the Lora and apply it on normal flux dev
The one without Lora is the full model, should be very slightly better quality but you have to download a full model again. But with quantization you can fit it in 8gb vram.
Lora is there so you don’t need to download a whole new flux model again, you just need to download the Lora and apply it on normal flux dev
The one without Lora is the full model, should be very slightly better quality but you have to download a full model again. But with quantization you can fit it in 8gb vram.
do these models/lora provide any benefit over regular flux 1 dev without utilizing the improved controlnet capabilities, just in regular generation or are they all purely meant to be used as tools?