【HiDream】TXT to IMG — v2.0

Use this HiDream workflow in ComfyUI to turn text prompts into images. It takes a HiDream model, generation settings, and LoRA inputs that can be toggled, and produces saved images with generation parameters embedded as metadata; SeedVR2 tiled upscaling can be enabled or disabled

The workflow includes prompt inputs, generation settings, LoRA handling, sampling, image saving, and a SeedVR2 upscaling stage.

The auto version includes automatic model downloads.

Three LoRA slots provide individual on/off toggles and strength controls, and additional LoRA modules can be connected.

SeedVR2 tiled upscaling can be toggled on or off, and saved images include generation parameters in metadata.

In ComfyUI, enable "Use Nodes V2 (Vue)", refresh the page, and then load the workflow; leaving it disabled may cause display problems.

Suggestion

Install ComfyUI-UmeAiRT-Toolkit through ComfyUI Manager by searching for UmeAiRT, or use the UmeAiRT Auto-Installer.

Suggestion

The listed model items are HIDREAM/Dev and seedvr2_ema_7b_sharp_fp8_e4m3fn_mixed_block35_fp16.safetensors.

The listed custom node is comfyui-umeairt-toolkit.

Manual model options are listed for 24 GB VRAM (base), 16 GB VRAM (Q8_0), and 12 GB VRAM (Q5_K_S).

For manual setup, place hidream_i1_dev_fp8.safetensors in ComfyUI\models\diffusion_models or hidream-i1-dev-QX_K_S.gguf in ComfyUI\models\unet; place ae.safetensors in ComfyUI\models\vae; and place clip_l_hidream.safetensors, clip_g_hidream.safetensors, t5xxl_fp8_e4m3fn_scaled.s,

Before using it, verify that the listing title is 【HiDream】TXT to IMG, the selected release is v2.0, and the listed files and custom node are available in ComfyUI.

Suggestion · non vérifié

The graph does not list node connections, so verify the wiring after loading.

Suggestion · non vérifié