Do I need Python or a GPU?
No. The hosted Space supplies the runtime and GPU; you only need a modern browser.
Zero-install identity editing
Krea2 Edit Without ComfyUI
Skip ComfyUI. Edit directly. Upload a source image, describe one change, and use the hosted Krea2 Identity Edit workflow from your browser.
krea2edit.app / live editor
The live editor runs in an embedded Hugging Face Space. If the shared GPU is busy, open it full-screen. Open the editor full-screen
ComfyUI vs browser — what's the difference?
Both routes can reach the same kind of identity edit. The browser optimizes for speed to first test; ComfyUI optimizes for repeatability and control.
| Browser editor | Local ComfyUI | |
|---|---|---|
| Setup | Open the page and upload an image | Install ComfyUI, nodes, model files, and dependencies |
| Hardware | Uses the hosted Space GPU | Uses your compatible local or rented GPU |
| Control | Focused controls exposed by the hosted app | Full graph, pinned weights, seeds, LoRA stacking, and batch workflows |
| Maintenance | Space owner maintains the environment | You maintain Python, nodes, models, and updates |
| Availability | Shared queue and uptime can vary | Available whenever your configured machine is ready |
| Best for | First tests, prompt validation, and occasional edits | Repeatable production, automation, and advanced tuning |
Workflow
How to get started in 30 seconds
You can reach the Generate step in about 30 seconds with a prepared image. Rendering time depends on the shared GPU queue.
- 01
Upload one clear image
Choose a sharp single-person photo you have permission to edit. - 02
Type one concrete instruction
Ask for one background, clothing, object, pose, or lighting change. - 03
Press Generate
Submit the job, wait for the shared GPU, then compare the result with your source.
FAQ
Krea2 without ComfyUI FAQ
What the hosted route removes—and what it does not.
Do I need to download the Krea 2 model or LoRA?
No. The hosted environment already contains its configured workflow and weights.
Is the browser output instant?
Submission is fast, but generation is not guaranteed to be instant. Shared GPU queues and Space availability affect wait time.
When should I move to ComfyUI?
Move local when you need pinned versions, repeatable seeds, batch generation, custom graphs, private infrastructure, or additional LoRAs.
No nodes. No Python. Just the edit.
Open the browser workflow now, or read the guide first for source-image and prompt recommendations.