# Check whether a 3D silhouette reads from its intended view [Read as HTML]() Use visual review when a model must read clearly in a particular output. State what the silhouette should communicate, from which view and at what size. A hypothetical character brief might require an arm-to-torso gap in the final frame. A screenshot shows the projected shape; modeling tools remain the source for geometry. ## Write the silhouette criterion before opening another view Make the brief testable: “At 1280 × 720, the arm and torso should remain separate in the three-quarter render.” This is an example, not a model we tested. Record the revision, render camera, output size, pose, and lighting. Judge the silhouette at final size; enlarging the viewport can make a tiny gap appear readable even when it disappears at the delivered size. A render attachment is enough for one review. Use a second view only to distinguish projection from form. ## Compare the intended view with a second view Inspect the arm-to-torso gap in the intended camera. If it vanishes, use another viewport angle to see whether that is projection overlap or a form that remains hard to read. This comparison concerns appearance, not mesh verification. In Blender, View → Perspective/Orthographic or Numpad 5 changes viewport projection, not the render camera. Orthographic apparent size is distance-independent; in perspective, farther objects appear smaller. Compare renders from the same render camera and output size. ## Choose a model revision or composition change Choose based on the deliverable. For a fixed-camera illustration, consider pose or form while keeping that camera fixed. If composition is still open, a camera change is another option. For a model that must read from several angles, check the required views before accepting a pose or form revision. Screenshots do not establish clearance, intersections, or dimensions as geometry facts. Change one choice and keep the other conditions fixed. If you change the camera, record both camera states instead of calling it a same-camera form comparison. The hypothetical record below shows the narrower case of a pose change with the camera held fixed. ```text Hypothetical silhouette review — no model test performed Brief: arm/torso separation at 1280 × 720 in the fixed three-quarter frame Baseline: outlines merge in that frame Second view: front viewport shows a visible gap; physical clearance remains unverified Chosen change: rotate the arm pose outward; keep camera, framing, output size, and lighting fixed Example result: gap is readable in the revised final-size frame Decision: accept for this stated view; check other required views and native geometry separately ``` ## Keep visual judgement and geometry checks separate A still supports discussion of silhouette, framing, contrast, and readability. It cannot validate topology, constraints, dimensions, collision, fit, or structure. Use Blender or CAD inspection, measurement, and simulation; keep requirements and exact values in text. Save this record with the render exports. A selected screen share helps only when the interactive viewport state is not already available as an image. ```text Criterion and intended view: [ ] Revision / render camera / output size: [ ] Baseline silhouette: [ ] Second-view observation: [ ] Form or presentation hypothesis: [ ] One change and same-camera result: [ ] Native geometry check, if needed: [ ] Decision and uncertainty: [ ] ``` ## Prompt to adapt > Hypothetical character render: at 1280 × 720, the arm and torso should remain separate in the three-quarter frame. Compare baseline and revision from the same render camera at final size. If the gap disappears, suggest a second viewport angle to separate projection from form. Do not claim geometry from pixels; I will choose a model or composition change and verify geometry in Blender. ## Source documentation - [Blender Manual: Perspective and Orthographic view]() ## Connect your source Add https://letitsee.com/mcp using remote MCP, Streamable HTTP, OAuth, and image results. Sign in with the same account and choose your source. [Share a screen or window with an MCP assistant]() cover setup and browser requirements. ## Set up without starting a capture [Set up your MCP client](https://letitsee.com/for-agents) with the full tool reference. The separate [capture page](https://letitsee.com/live) can replace the active capture page for your account. It does not take a photo by itself. ## Related guides - [Learn CAD sketch constraints by predicting how a line changes]() - [Review a clipped interface label at a fixed viewport]() [All camera and screen guides](https://letitsee.com/guides)