Frame the camera
Look through the scene camera (Numpad 0) and frame the shot. Keep the camera level for straight verticals.
OpenAI · GPT Image 2.5 × Blender
Turn Blender viewport shots into photoreal renders with GPT Image 2.5 in RenderChat. Upload your Blender viewport render, describe what you want, and refine the render in chat.


Drag to compare the Blender viewport render with the GPT Image 2.5 render. Input is an AI-generated example.
Before you start
Cycles can render anything, but lighting, materials and sample times add up during design. Grab a fast Workbench or viewport image, describe the look you are after and develop it in RenderChat, then decide which direction deserves a full Blender render.
Look through the scene camera (Numpad 0) and frame the shot. Keep the camera level for straight verticals.
Solid shading with Cavity and Shadow enabled gives the AI form and light cues without distracting colors.
View ▸ Viewport Render Image, or render with the Workbench engine, and save a PNG 2000 px or wider.
How it works
by OpenAI
Frame the camera, use clean solid shading and export the image. The checklist above has the details.
The start button on this page opens a new chat with GPT Image 2.5 Sunburst selected. You can also pick it from the model menu in the composer.
Attach the image and say what must stay, then the materials, light, setting and people. Start from an example brief below.
Ask for changes in follow-up messages. GPT Image 2.5 continues the same conversation, so a short instruction such as "warmer light, keep everything else" is enough.
Download the image at the resolution you chose (1K).
Example briefs
Copy a brief, paste it into RenderChat with your Blender viewport render, and change the materials and mood to match your project.
Photoreal interior render of this Blender scene. Keep the furniture layout and camera. Wide-plank oak floor, white plaster walls, boucle armchair, large window with soft morning light, a few books and plants.
Photoreal exterior render of this Blender model. Keep the form and camera. Charred timber cladding, black window frames, forest setting with mist, early morning light.
GPT Image 2.5 in RenderChat
OpenAI's GPT Image 2.5, offered in RenderChat as two variants, Sunburst and Flare. Both run as native conversations, so every follow-up message keeps working on the same image thread. Sunburst is RenderChat's default model.
Quality check
Compare
RenderChat lets you switch models in the same chat. Run your Blender viewport render through two or three and keep the result that respects your design best.
| Model | Maker | Output | References | Follow-ups | Plan | Guide |
|---|---|---|---|---|---|---|
| GPT Image 2.5This page | OpenAI | 1K | Supported | Continues the chat | Free | |
| Nano Banana Pro | 1K · 2K · 4K | Up to 14 images | Continues the chat | Pro | ||
| Nano Banana 2.1 | 1K · 2K · 4K | Up to 14 images | Continues the chat | Free | ||
| FLUX.2 Pro | Black Forest Labs | 1K | Up to 9 images | Fresh request | Pro | |
| Seedream 5.0 | ByteDance | 1K · 2K | Up to 10 images | Fresh request | Pro | |
| Kling O3 | Kuaishou | 1K | Up to 10 images | Fresh request | Pro | |
| Reve 2.1 | Reve | 4K | Up to 8 images | Fresh request | Pro | |
| Krea 2 | Krea | 2K | Up to 10 images | Fresh request | Pro | |
| Ideogram 4 | Ideogram | 1K · 2K · 4K | Up to 1 image | Fresh request | Pro | |
| GPT Image 2 | OpenAI | 1K | Supported | Continues the chat | Pro | |
| Nano Banana 2 Lite | 1K | Up to 14 images | Continues the chat | Pro |
More with GPT Image 2.5
FAQ
Yes. GPT Image 2.5 is available on the Free plan as well as Pro.
GPT Image 2.5 renders at 1K in RenderChat, with auto + 10 fixed ratios.
Yes. GPT Image 2.5 runs as a native conversation in RenderChat, so each follow-up continues from the previous turn. Short edit instructions work well.
Use the start button on this page — it opens a new chat with GPT Image 2.5 Sunburst selected — or choose it from the model menu in the chat composer.
Do, when the image is final. RenderChat is for the stage before that: testing materials, light and mood quickly without setting up shaders and lights for each idea.
Either. Plain solid shading gives the AI the most freedom; viewport materials steer it toward colors you already chose.
Open RenderChat with GPT Image 2.5 selected, upload your Blender viewport render and start from one of the briefs above.
Cookies help us improve RenderChat. Accept to allow analytics and advertising cookies (Google, Meta), or decline to keep them off. Privacy