Render Settings, Resolution, and Color Management Guide

Define the deliverable before the renderer

A final image, comp plate, editorial preview, and texture bake require different settings. Write down pixel dimensions, aspect ratio, frame range, frame rate, alpha requirement, channels, naming, bit depth, and destination. Confirm whether the client expects display-ready pixels or scene-linear plates. This prevents a technically clean render from being rejected because it has the wrong crop, transfer function, premultiplication, or missing AOVs.

A repeatable preflight workflow

Open the approved scene version, resolve missing assets, set the production camera, and load the sanctioned render preset. Render a diagnostic frame with a gray reference, saturated colors, highlights, transparency, motion blur, and depth. Inspect it in both the DCC viewer and the downstream compositor. Then submit a small first-middle-last range. Only release the full frame range when dimensions, channels, color interpretation, and temporal effects match.

Resolution, sampling, and output settings

Set width and height explicitly rather than relying on a percentage control. Verify pixel aspect is normally 1.0 and confirm overscan or crop windows. Choose samples by measured noise in difficult regions, not an arbitrary maximum. Use denoising only after checking detail retention. Save layered OpenEXR for compositing when required; use PNG or another display format for review copies. Keep compression lossless where pixel accuracy matters.

Color management boundaries

Textures tagged as color should normally receive an input transform; data maps such as normals, roughness, depth, and masks must bypass color conversion. Lighting and shading calculations should remain scene-linear. A view transform maps that data to a display appearance but should not be silently baked into plates intended for compositing. Record the configuration, working space, display, view, look, and exposure. Screenshots are not reliable color evidence without these settings.

Decision example: film plate or web still

For a composited shot, choose scene-linear OpenEXR, sufficient bit depth, unpremultiplied or agreed alpha handling, and required AOVs; provide a separate display-referred preview. For a web still, render through the approved view, convert to the delivery color space, embed the expected profile, and inspect at target size. The lighting scene can be identical, but output transforms and file formats differ because the next consumer differs.

Failure diagnosis by symptom

A washed-out image usually indicates a view transform applied twice or a linear file read as display-referred. An overly dark image can be the inverse mismatch. Jagged edges may come from low samples, scaling after render, or an incorrect pixel aspect. Missing transparency points to output format, alpha mode, or holdout configuration. Flickering denoise or fireflies require temporal sample analysis. Compare raw values and metadata before changing lights.

Evidence checklist for final frames

Record renderer and version, resolution, pixel aspect, camera, frame range, sample controls, denoiser, motion blur, and output path pattern. Include the color configuration identifier plus input, working, display, and view transforms. Inspect several frames at 100 percent, verify alpha and every required AOV, and compare hashes or file counts against the submission list. Save one approved reference image so farm reruns can be checked against a known target.

Related references and next step