Tripo Studio 集成 OpenAI GPT-Image-2.5 Sunburst，打通图像参考到 3D 资产工作流

Friday, October 16th, 2026 Article by Tripo AI

[Sponsored] Tripo explains how integrating OpenAI’s new GPT-Image-2.5 Sunburst model into Tripo Studio has improved workflow when generating 3D assets from reference images by giving artists a way to solve image problems before they become geometry problems.

Tripo has integrated GPT-Image-2.5 Sunburst into Tripo Studio, giving artists a more connected workflow for creating image references and turning them into 3D assets.

The update brings image generation, targeted image editing, multi-view generation and 3D model creation into the same browser-based workspace. Instead of beginning with a single prompt and transferring files between several applications, users can develop a concept, adapt it for reconstruction and send it directly into Tripo’s Image to 3D Model workflow.

From image generation to 3D-ready reference
Tripo Studio’s Generate Image feature is intended to address the gap between creating an appealing concept and preparing one that can be reconstructed effectively. Artists can start with a text prompt, upload an existing image, or combine text and an image. They can then choose a task-specific template rather than relying on a blank prompt to produce a suitable reference.

The current template set includes Character Extraction, Head Extraction, T Pose, Character Completion, Variants, 3D Enhance, 3D Print-Ready, Photo to Figure and Asset Extraction. Each option uses a preconfigured prompt tuned for a particular downstream task.

Character Extraction separates elements like the head, body and accessories so that they can be generated as individual 3D assets. Head Extraction prioritises facial detail, while Character Completion restores body parts that are missing from a cropped or partial source image. T Pose converts a character into a standard presentation better suited to rigging and animation.

The other templates cover design exploration and asset production. Variants creates related designs with a consistent visual direction; 3D Enhance strengthens depth cues; and 3D Print-Ready produces a high-contrast, white-model-style reference intended to make the geometry easier to interpret. Photo to Figure turns a portrait or full-body photograph into a stylized figurine concept, while Asset Extraction separates props and environment elements from a larger scene.

Edit images and generate additional views
Once an image has been generated, users can review it inside Tripo Studio and make focused revisions with Edit Image. The tool accepts text- and reference-guided changes, enabling an artist to adjust a specific part of the design while retaining more of the surrounding subject and composition. This makes it possible to correct an unsuitable pose, clarify a silhouette or revise an accessory before committing credits and time to the 3D stage.

For subjects that cannot be described adequately by one view, Generate Multi-Views creates front, left, right and back images. These views expose structural information that may be absent from the initial picture, including the side profile, rear design and asymmetrical details. Tripo can generate a model from either a single image or multi-view references, so users can choose a faster concept workflow or provide more visual evidence for a complex asset.

Cutout can isolate individual elements before generation. Batch to 3D can then process up to 10 prepared images at once, which is useful when converting a set of character parts, props or environment pieces rather than a single hero asset. The resulting models appear in Tripo’s asset panel, where they can be previewed and taken into later processing.

A connected workflow for characters, games and 3D printing
For character work, an artist can generate or upload a concept, use T Pose to standardize the stance, create the 3D model, and continue with Tripo’s Smart Topology Mesh, rigging and animation tools. Smart Topology Mesh generates editable topology, including native quad meshes, and lets users set different polygon targets before exporting the asset for further work.

For game environments, Asset Extraction and Cutout can break a scene concept into separate objects. The selected images can be converted in a batch, then transferred to a DCC application for scene assembly. Tripo’s DCC Bridge supports a more direct hand-off to content-creation software, reducing the need to download and re-import every iteration manually.

The workflow can also be used to create coordinated asset collections. A designer can run one reference through Variants, isolate the preferred results and batch-generate a family of toys, props or decorative objects that share a common style. For physical output, 3D Print-Ready or Photo to Figure can be combined with multi-view generation and Tripo’s high-detail modelling options before the mesh is inspected and exported for print preparation.

Artists who do not have a source image can begin with Tripo’s Text to 3D Model workflow. Tripo first turns the description into a reference image, which can be iterated with the same GPT-Image-2.5 Sunburst-powered image tools before being converted to 3D. 3D assets can be segmented, retopologized, textured, rigged or animated, depending on the intended use.

Conclusion
By connecting concept generation, reference preparation and 3D creation, the latest update to Tripo Studio is intended to shorten the route from initial idea to a model that can be refined for games, animation, visualisation or fabrication. More importantly, it lets artists solve reference-image problems before they become geometry problems: they can expose hidden details, separate assets, standardize a pose and approve the design without leaving Tripo Studio.

Use Tripo Studio to generate 3D models from text and images