One canvas for everything you do with Krea 2 Turbo. Type a rough idea, let the built-in prompt enhancer write the full prompt, and get pictures in every shape. Regional prompting for up to four characters, detailed and simple upscaling, inpainting, image to image and overnight batches from a plain text file are all in the same file. Every part sits in a colour-coded group with a short note beside it that tells you what to click.
FULL GUIDE HERE!
Why you will like it
- A prompt enhancer that is just a switch. Type tags, plain words or a half sentence. The enhancer rewrites it into a full descriptive prompt using a small text model. Turn it off and your words go through exactly as typed. A preview box shows the prompt that is really sent.
- The Brain. One prompt box, the seed, steps and CFG in one place. Everything else reads from it.
- Every shape, draft then hi-res. Wide, tall, ultra wide and square. Try a prompt in a small fast DRAFT, then switch on the HI-RES group under it to enlarge and refine the keeper with the same seed. Or use a ONE PASS group for a finished picture in one go.
- Regional prompting. Up to four characters side by side. Write one prompt with BREAK before each person and the 2, 3 or 4 region mask is picked for you.
- Five LoRA setups. Pick the one you want with one panel.
- Overnight batches. Put prompts in a text file, split them with a line of =====, turn the prompt file on and queue. A start-at-a-set-time node holds the first picture until the hour you choose.
- Notes right next to what they explain. Short, plain, no meta chatter.
- Four files, pick your size. Pictures only, pictures + Wan, pictures + MiniMax H3, or video only. The Wan and H3 versions add a simple 3-part chain to extend a video. The video-only file has the full long chain: up to 10 scenes for Wan and H3, each H3 scene can start from and end on its own picture, and extra LoRA slots.
- Clean copies for posting. Hover a picture and press C. A copy is saved with the prompt hidden and the embedded workflow removed. Your model, LoRAs and seed stay, so the Resources box still fills in on Civitai.
What is inside
| Group | Key | What it does |
|---|---|---|
| A1 The Brain | 2 | Prompt, seed, steps, CFG, and every panel of switches. |
| LoRA setups 1 to 5 | 1 | Five LoRA boxes. Switch on the one you want. |
| Enhancer | 7 | Rough idea or tags into a full Krea 2 prompt. On or off with one switch. |
| ONE PASS Wide, Tall, Ultra, Box | 3 | One finished picture per group. Turn on any number. |
| A2 DRAFT and B HI-RES | = | Small fast draft of each shape, and the hi-res group under it that enlarges and refines it. |
| H1 / H3 Regional prompting | 4 and 6 | Up to 4 characters side by side. |
| ONE PASS Regions | minus | The picture groups that go with H1 and H3. |
| D Upscale (detailed) | 8 | Tiled 2x upscale that adds fresh detail. |
| E Inpainting | 9 | Paint a mask, write what goes in it. |
| F Simple upscale | 0 | Fast model upscale, no new detail. |
| G Image to image | 5 | Restyle or change a picture. |
| J Batch final upscale | [ | Upscale a whole folder of keepers, sized by shape. |
| K Prompt file | | Overnight batch from a text file. |
| Video files | | 2_Krea2_Brain_Images_plus_Wan.json (Wan), 1_Krea2_Brain_Images_plus_MiniMax_H3.json (MiniMax H3 with sound), and 4_Video_Only_Wan_and_H3.json (both, with the 10-scene long chain). |
| S Start at a set time | | Queue before bed, the first picture waits for the hour you set. |
Quick start (5 steps)
1. Install the custom nodes listed below (ComfyUI-Manager makes it one click each).
2. Copy the anima_batch_loader folder from the download into ComfyUI/custom_nodes/, then copy everything in the download's input folder into ComfyUI/input/. Restart ComfyUI.
3. Put the model files in their folders (list below).
4. Drag the workflow JSON onto ComfyUI. Pick your files in the model boxes.
5. Press 2, type a prompt in the positive box and press Run (the Wide group starts on).
Requirements
Custom nodes (install with ComfyUI-Manager):
- rgthree-comfy
- ComfyUI-Lora-Manager
- ComfyUI_UltimateSDUpscale
- ComfyUI-ppm, for regional prompting
- ComfyUI-GGUF, only for the Wan video groups
- the anima_batch_loader folder from the download
You need a ComfyUI recent enough to support Krea 2, so update it first.
Models (from [Comfy-Org/Krea-2](https://huggingface.co/Comfy-Org/Krea-2)):
| What | Folder |
|---|---|
| krea2_turbo_fp8_scaled.safetensors | models/diffusion_models |
| qwen3vl_4b_fp8_scaled.safetensors | models/text_encoders |
| qwen_image_vae.safetensors | models/vae |
| 4x_foolhardy_Remacri.pth (any 2x or 4x upscale model works) | models/upscale_models |
Optional, for the enhancer: qwen3.5_9b_int8_convrot.safetensors from [Comfy-Org/MOSS-Image](https://huggingface.co/Comfy-Org/MOSS-Image) into models/text_encoders. No enhancer file? Set its switch to OFF.
Optional, for video: the Wan 2.2 GGUF models and lightning LoRAs, and the MiniMax H3 model files. The file names and folders are listed in the guide, section 4.
## Settings that work
Krea 2 Turbo: 8 to 12 steps, CFG 1. At CFG 1 the negative prompt does nothing. Write short or long sentences, not tag lists (or leave the enhancer on and type tags).
Regional prompting
Turn on H1 and ONE PASS REGIONS 16:9 (one switch in the Brain), and turn the enhancer off. Write one prompt with the word BREAK before each person:
two friends on a train platform at dusk BREAK left girl: a young woman with a brown ponytail and a blue jacket BREAK right boy: a tall man with short black hair and a gray coat
The 2, 3 or 4 region mask picture is chosen for you. No BREAK? Then the four region boxes in the Brain are used instead. Masks for 2, 3 and 4 characters, wide and tall, are in the download.
Video
Pick the file that fits: 3_Krea2_Brain_Images_Only.json (pictures only), 2_Krea2_Brain_Images_plus_Wan.json (pictures + Wan), 1_Krea2_Brain_Images_plus_MiniMax_H3.json (pictures + MiniMax H3) or 4_Video_Only_Wan_and_H3.json (video only, both models). In a video file, load any picture you made in the Video Controls box, turn on one video switch (V2 for Wan, V4 and V4a for H3) and press Run. To extend a video turn on the chain group (V2x or V4x) and write what happens next in the Chain Part boxes. The video-only file goes to 10 scenes: pick how many in the VC2 panel and write PART 2: to PART 10: in your prompt. Each scene is another full run, so more scenes take longer.
Notes
- The upscale, inpaint and image to image groups share the same Turbo sampler settings as the picture groups. For more change in image to image or inpaint, raise the denoise on its sampler.
- Leave your computer awake for overnight runs, and ComfyUI open.
The full guide, with every group explained step by step, is in the article.