update workflow to correct version

add ComfyUI workflow (t2i + SeedVR2 upscale)

Rename fullgreed_int8comfy.safetensors to fullgreed_int8.safetensors

Rename fullgreed_i8_plain_comfy.safetensors to fullgreed_int8comfy.safetensors

Delete flux_1_vae_f16.safetensors

add money_vae_f16 (16ch VAE)

Delete fullgreed_fp8_e4m3fn.safetensors

Delete fullgreed_i8v2_comfy.safetensors

Delete fullgreed_i8_comfy.safetensors

Delete qwen3vl_4b_bf16.safetensors

Delete fullgreed_i8v2_refiner_lora_r64.safetensors

Upload fullgreed_i8_plain_comfy.safetensors with huggingface_hub

Upload fullgreed_i8v2_refiner_lora_r64.safetensors with huggingface_hub

Upload fullgreed_i8v2_comfy.safetensors with huggingface_hub

Mirror Comfy-Org Qwen3-VL-4B bf16 text encoder (same model Draw Things pairs with fullgreed)

Mirror Comfy-Org qwen_3_4b text encoder (Z-Image standard) for one-stop import

Add Flux 16ch VAE (verified identical to Draw Things flux_1_vae / money_vae)

Add INT8 (ConvRot) ComfyUI quant of fullgreed

Add Colab int8 (ConvRot) quantization pipeline script

Add fp8_e4m3fn quant (Z-Image, all 453 tensors F8_E4M3)

README: correct architecture to Z-Image + ComfyUI usage

Fix keys to Z-Image layout (adaLN_modulation.0); matches John2386/greed 453/453

Upload fullgreed_f16 (NextDiT/Lumina) diffusion model, fp16

Add model card

initial commit