> For the complete documentation index, see [llms.txt](https://docs.m-xr.com/llms.txt). Markdown versions of documentation pages are available by appending `.md` to page URLs; this page is available as [Markdown](https://docs.m-xr.com/marso-studio/reference/nodes-reference.md). # Nodes reference Marso Studio is built around **nodes**. Each node is one step — generate a scene, break it into objects, generate or load an asset, generate or edit an image, apply materials. Connect a node's output to the next node's input to build a pipeline. When you create a project you pick a **workflow**, and the starting nodes are placed for you: | Workflow | You start with | | -------------- | ----------------------------------------------- | | **Bulk** | **Scene Generation** → **Detect Objects** | | **Individual** | **3D Model Upload** → **Generate PBR Textures** | Nodes are grouped into **Scene**, **Image**, **3D** and **Utility**. ## Scene ### Scene Generation Generate a scene image from a text prompt — a whole environment rather than a single object. This is the starting node in a Bulk project. * **Input:** text prompt * **Output:** scene image * **Cost:** 10 credits ### Detect Objects Break a scene image into its individual objects. Returns a list of detected objects with a preview, which you then review and confirm before any generation runs. Everything is selected by default. * **Input:** an image of a scene, and optionally the object categories to look for * **Output:** a set of detected objects * **Cost:** 20 credits per run You can run it again with different categories if the decomposition isn't right — each run costs the same 20 credits. See [Detecting & selecting objects](/marso-studio/scenes/detecting-and-selecting-objects.md). ### Generate Reference Images Generate a clean reference image for each object you selected, ready for 3D reconstruction. * **Input:** the confirmed object selection * **Output:** one reference image per object * **Cost:** 10 credits per object > Each object needs to fill most of its reference image for 3D generation to succeed. Objects that were small or heavily occluded in the scene are the usual failures. ## Image ### Image Upload Upload an image to use as a starting point in your pipeline. * **Input:** image file * **Output:** image * **Cost:** no credits ### Image Generation Generate an image from a text prompt. * **Input:** text prompt * **Output:** image * **Cost:** 10 credits ### Multiview Generate a multi-view set of an object from a single image — useful for a stronger 3D reconstruction. * **Input:** image * **Output:** multiple views of the object * **Cost:** 2 credits Only **Hunyuan 3.1** accepts multi-view input, so this node feeds a Hunyuan generation. See [Choosing a generator](#choosing-a-generator). ### Image Edit Edit or refine an existing image with a prompt. * **Input:** image + prompt * **Output:** edited image * **Cost:** 10 credits ## 3D ### 3D Model Upload Upload your own 3D asset (`.glb`, `.usdz` or `.fbx`). See [3D file support](/marso-studio/preparing-your-assets/3d-file-support.md). This is the starting node in an Individual project. * **Input:** 3D file * **Output:** 3D model * **Cost:** no credits ### Generate 3D Model Turn an image into a 3D mesh. The object should fill most of the image — sparse or distant subjects often fail. * **Input:** an image, or a multi-view set * **Output:** 3D model * **Cost:** 30–120 credits, depending on which generator you pick #### Choosing a generator Reconstruction is the one stage where no single model wins on every subject, so **Generate 3D Model** lets you pick which one runs. **They are not all priced the same:** | Generator | Single image | Custom face count | Multi-view input | | --------------------------- | ------------ | --------------------- | ---------------- | | **Tripo H3.1** | **30** | 30 — no extra charge | Not supported | | **Hunyuan 3.1** *(default)* | **40** | 55 *(+15)* | **60** | | **Meshy 6** | **80** | 80 — no extra charge | Not supported | | **Meshy 7** | **120** | 120 — no extra charge | Not supported | Three things worth taking from that table: * **Tripo H3.1 is the cheapest option** — and cheaper than the default. * **Meshy 7 costs four times as much as Tripo H3.1.** On one asset that's a rounding error. Across a twenty-object scene it's the difference between 1,430 credits and 3,230, so **check which generator is selected before you run anything at scale** — see [Credits & billing](/marso-studio/reference/credits-and-billing.md). * **Only Hunyuan 3.1 takes multi-view input.** The **Multiview** node feeds Hunyuan generations. Hard-surface props, organic shapes and busy reference images all favour different models, and the differences are hard to predict from the image alone. **If a mesh comes back wrong, re-running the same reference image through a different generator is usually a better move than changing the image** — just check what the switch costs first if you're working at scene scale. Switching generator re-runs the generation, charged at that generator's rate. Whichever you pick, the mesh feeds the same **Generate PBR Textures** stage afterwards. The generator decides your geometry and base colour; I2M decides your materials. **Custom face count** — you can set a target polycount instead of taking the default. It's offered on every generator, but only **Hunyuan 3.1 charges for it**: +15 credits, so 55 in total, and on Hunyuan it's single-image only, not multi-view. On Tripo H3.1, Meshy 6 and Meshy 7 a custom face count is included in the base price. ## Utility ### Generate PBR Textures Run I2M to predict a full PBR material stack — **albedo, metallic, roughness and IOR** — for a textured mesh. * **Input:** textured 3D model * **Output:** textured 3D model + individual PBR maps * **Cost:** 30 credits Generated meshes are compressed automatically before texturing. Compression is free and needs no node. ## Credit costs at a glance | Node | Credits | | ------------------------------------------------- | --------------------------------------- | | Scene Generation | 10 | | Detect Objects | 20 per run | | Generate Reference Images | 10 per object | | Image Upload | — | | Image Generation | 10 | | Image Edit | 10 | | Multiview | 2 | | 3D Model Upload | — | | Generate 3D Model (single image) | 30–120, by generator | | Generate 3D Model (multi-view — Hunyuan 3.1 only) | 60 | | Generate 3D Model — custom face count | +15 on Hunyuan 3.1 · free on the others | | Generate PBR Textures | 30 | Your choice of 3D generator is the biggest single variable in what a run costs — see [Choosing a generator](#choosing-a-generator) above, and [Credits & billing](/marso-studio/reference/credits-and-billing.md) for how credits work and how to budget a scene. --- # Agent Instructions This documentation is published with GitBook. GitBook is the documentation platform designed so that both humans and AI agents can read, navigate, and reason over technical content effectively. Learn more at gitbook.com. ## Querying This Documentation If you need additional information that is not directly available in this page, you can query the documentation dynamically by asking a question. Perform an HTTP GET request on the current page URL with the `ask` query parameter, and the optional `goal` query parameter: ``` GET https://docs.m-xr.com/marso-studio/reference/nodes-reference.md?ask=&goal= ``` `ask` is the immediate question: it should be specific, self-contained, and written in natural language. `goal` is optional and describes the broader end goal you are ultimately trying to accomplish on behalf of the user. GitBook uses it to tailor the answer towards what is most useful for that goal. The response will contain a direct answer to the question and relevant excerpts and sources from the documentation. Use this mechanism when the answer is not explicitly present in the current page, you need clarification or additional context, or you want to retrieve related documentation sections.