Seedream 5.0 Pro: Image Edit
Edit images with precision using Seedream 5.0 Pro, targeting specific regions while preserving lighting, depth, and texture. The workflow takes 1 input image and produces 2 output images through an optional Painter node and preview. Ideal for product photography edits, portrait consistency adjustments, and creating structured layouts with multilingual text.
简介
Seedream 5.0 Pro: Image Edit is a precision image-to-image and inpainting workflow that lets you change exactly what you want while keeping the original lighting, depth, and texture intact. It starts with LoadImage to bring in your source photo, then optionally uses Painter to create a clean mask for targeted edits. The core transformation happens in ByteDanceSeedreamNodeV2, which conditions on your input image (and mask if provided) to apply prompt-driven changes without destroying structure. Finally, SaveImageAdvanced writes out clean results with controllable filenames and folders.
你需要提供
- 正在读取工作流
在 ComfyUI 中运行。 此工作流无法在本页面运行。你可以在 Comfy Cloud 中打开(已预先加载所需内容),或下载后在自己的 ComfyUI 中使用。
在 Cloud 中试用常见问答
It’s the core image-to-image/inpainting model node. It conditions on your input image (and optional mask) plus your prompt to apply localized, structure-preserving edits.
Use the Painter node to create a mask: paint white where edits should occur and black where they should not. Feed this mask into ByteDanceSeedreamNodeV2 so only the painted region is modified.
Use the same prompt, negative prompt, seed, steps, guidance/CFG, and keep denoise strength low to moderate. Consistent framing and lighting in your source images also help Seedream preserve structure.
The first is the main edited result from ByteDanceSeedreamNodeV2. The second is a Painter-guided preview that mirrors the masked edit for quick comparison. Use the main edited image for final delivery, and the preview for QA.