Consistent colors, from Blender to the delivered file
To find your image in another software, pass on its color interpretation along with its pixels: file space, configuration, view and expected output. Then compare the same export in both applications. Correcting the material before this verification can mask a reading error. This method aims for a controlled exchange; it does not guarantee an identical appearance on all screens.
Decide whether you are delivering an image to look at or to work on
A visual intended for a web page and a render intended for compositing don't serve the same purpose. The first carries an approved look. The second must allow you to adjust exposure, balance and effects before output. Ask the recipient which one they expect, in which application and with which version. A file named "final" answers none of those questions.
Blender distinguishes between the spaces needed for computation, storage and display. Rendering and compositing rely on linear scene values; presenting them on screen requires an adapted interpretation. "Linear" alone isn't enough to name a space: also note its primaries, for example Linear Rec.709 or ACEScg if that's actually your configuration. The color spaces manual explains this separation.
Settle the deliverable before the artistic tweaks. If all you receive is "send an EXR", have them specify the expected space and the display treatment. The extension alone doesn't constitute a color agreement.
| Need | Agreed file | Priority check |
|---|---|---|
| Approve a look | Review image with documented color output | Same contrast, same hues, same framing at the start |
| Continue compositing | EXR with named linear space and view settings | Input correctly interpreted before any correction |
| Put both back | Working master and visual reference identified separately | The reference makes it possible to reconstruct the master's appearance |
Note the five decisions that change how it reads
Document the Blender version and the recipient's version, then the file space, the OCIO configuration, the viewing settings and the intended output. For viewing, record the view, any look, the exposure and any additional curves. Photographing only the view menu leaves too much implicit. A short text note accompanies successive exchanges better.
In the receiving software, distinguish three operations: declaring what the input contains, choosing the space to work in, then choosing how to view and export it. Assigning another space to existing pixels is not the same as converting those pixels to that space. Avoid changing both operations at once during diagnosis.
After Effects lets you set the media, project, display and output module separately; its OCIO engine accepts a custom configuration. These possibilities depend on the version used. Adobe's documentation helps you find these four points. An option with the same name in two applications does not prove that their configurations are identical.
Building a comparison that isolates the problem
Choose a representative image with a readable shadow, a neutral surface, a saturated hue and a bright highlight. Keep the framing and delivery resolution. Temporarily disable the corrections added by the recipient and keep a copy of their project before making any changes.
Export a working master and a viewing reference from the same image. In Blender, Save As Render applies render management according to the format: view and display transform for a presentation image such as PNG, render output space for an intermediate such as OpenEXR. Check your save settings instead of assuming that two buttons produce the same file. The save manual describes this distinction.
Open the master on the recipient's side, declare its actual space, then align the viewing. Compare the broad masses first, then the details at full size. If you use a difference between two images, compare files brought into the same space and the same framing. A difference between a screenshot and a raw EXR does not diagnose the fidelity of the transfer.
Example: an orange ceramic piece for a campaign visual
Here is an illustrative example of preparation for Blender 5.0, with no measured render. Menus and configurations should be verified in the versions actually used. A scene shows an orange mug, a cream label and a nearly white highlight. Compositing should add a title and a graphic shadow, then deliver a viewing image. The creator provides a working EXR and a reference PNG, with the space, view and exposure noted from their scene.
When the EXR is opened, the mug looks darker than in the PNG. The first action is to check the input interpretation, then the viewing transform. The material does not change. When those settings match, the label, the highlight and the mug are compared separately: correct neutrality with different saturation may point to the view or a remaining correction; a general shift may indicate a conversion applied incorrectly. These are diagnostic leads, not infallible signatures.
The title is then added in the correctly interpreted project. A new output is reopened as a standalone file. The agreement concerns that output, not just the software window. If the recipient cannot reproduce the configuration, the decision is made explicitly to work from an appearance that is already transformed, with less latitude for recovering the highlights. We do not rename that file "linear master".
Fix the pipeline before you fix the picture
Does a render look washed out after a transformation applied twice? Find the step that applies it and remove the repetition; don't immediately compensate with an approximate inverse curve. Does an export lose its highlights while the master retains them? Examine the format and the output. Do identical RGB numbers accompany two different appearances? Check their space and viewing conditions.
To diagnose a discrepancy, change a single step, keep both states, and note what it resolves. A correction that improves the orange while degrading the gray is not validated. If only the translucent edges differ, continue with alpha compositing, because the background color can skew your judgment.
OpenColorIO distinguishes transformations between spaces from those intended for display. Its configuration validation tool checks the structure, without certifying the artistic accuracy of the transformations. Consistency must therefore also be judged on your image. No additional tools are required here: start with the settings and viewers already available.
Accept the file in its distribution context
Reopen the delivered file in a clean session and verify that the announced profile or space is preserved. Note the exact file, the application version and the viewing conditions used for the agreement. For a series, also check a dark image and a very bright image; a single neutral image can hide a range problem.
The screen, its profile and the playback software remain the limits of the comparison. Adobe in particular distinguishes display conversion from project data. A validation on your workstation does not oblige every phone to display the same shade. For a critical brand color, agree on a medium and a validation reference rather than promising universal equality.
Once the chain is accepted, keep this sheet with the elements provided in the project export. That guide organizes the overall handoff; this check establishes how to correctly read the images it contains.
Your questions.
Why does my EXR look different from the Blender render?+
First check the input space and the view transform. A working EXR and a presentation image can contain different representations of the same render. Align their interpretation before modifying materials.
Should I apply the view twice to get the contrast back?+
No. Look for where the view is already applied: viewport, compositing effect or export. Applying it an extra time can temporarily hide a bad input setting and complicate the output.
Is a single OCIO configuration enough?+
It provides a common framework, but the selected color spaces, view transform, exposure, effects, and output must also match. Check a representative file in the versions actually being used.
Can I guarantee color on every screen?+
No. Define a suitable output and validation conditions. Screen capabilities, its profile, and the player still influence the appearance.