An HDR video finish can give bright highlights and saturated color more room, but it cannot rescue a weak shot. Runway released Ruby on August 20, 2026 as an SDR-to-HDR color-grading model. For an APOB creator, the practical question is smaller: does one already-approved hero clip gain enough on an HDR-capable display to justify another paid render, a second delivery file, and extra review?

Test One APOB Clip Before Committing to HDR

This workflow keeps generation and finishing separate. Build the strongest SDR master with the APOB AI Video Generator, freeze it, then test that exact file in Ruby. APOB AI prepared this guide from the official Runway release and pricing pages on August 28, 2026. It reports a controlled method, not a private benchmark or a promise that HDR improves every clip.

This SDR to HDR protocol publishes no invented result. It uses a two-attempt ceiling, rejects every unsupported claim, and keeps a known-good SDR fallback.

SDR presumption

Begin with an SDR presumption: keep the approved SDR master unless the destination, source file, and controlled comparison all justify an HDR companion. Ruby creates a new finishing option; it does not change what happened during generation or make every player, platform, and display HDR-ready.

Release facts. Runway's August 20 changelog entry describes Ruby as a native color-grading model that transforms SDR video into HDR. The entry says it can update brightness and color range and is available through Tool Mode, Workflows, and Runway Dev. Treat those as vendor-stated capabilities. The evidence date is August 28, 2026; recheck the changelog before publication.

Supported inputs. The same release says Ruby accepts outputs from Runway models, third-party models, and existing video files. That makes a controlled APOB-first handoff possible: generate or animate an owned asset, download one untouched master, and give Ruby that file without changing the prompt, cut, or duration. Compatibility is not proof of improvement; it only establishes that the test route exists.

Access and cost. Runway lists Ruby for Pro+ plans and its current models and credits page shows 20 credits per second. A ten-second input therefore has a listed generation estimate of 200 credits before any failed attempt or repeat. Save the dated page, account surface, actual task cost, and receipt. Do not turn one account observation into a permanent price claim.

Display-chain proof

Do not send a whole campaign through an HDR video converter because a label sounds premium. Select a candidate only when its delivery context and visual content create a plausible benefit. The decision begins before credits are spent.

HDR versus 4K. HDR describes a wider brightness and color range; 4K describes pixel dimensions. A 4K HDR video combines both, but one does not guarantee the other. Upscaling may add pixels without restoring clipped highlights, while an HDR grade may change range without increasing resolution. Record resolution, codec, bit depth, and display mode separately so reviewers do not credit the wrong process.

Candidate clips. Choose one short hero clip with deliberate highlights, gradients, saturated objects, skin tones, or a product surface that could reveal grading problems. Prefer an approved clip made from owned or licensed assets. Reject a candidate with broken anatomy, unreadable product text, severe compression, or a failed narrative beat; those defects belong upstream, not in HDR video editing.

Score the source before conversion: identity, motion, product fidelity, composition, exposure, color, text, and export integrity. Each criterion is pass or fail. Ruby enters the workflow only if the source passes every protected criterion and the destination can show HDR.

When SDR is enough. Keep SDR when the destination does not reliably support HDR, the client requires one simple master, the source has little highlight or color-range pressure, or the comparison shows no meaningful gain. SDR is also the fallback when faces become unnatural, highlights look harsh, gradients band, or platform conversion is uncertain. “No conversion” is a valid production decision, not a failed experiment.

Master-file freeze

APOB should remain the creative source of truth. Its Runway-compatible video workflow helps a creator develop the subject, motion, duration, and approved SDR clip before an external finishing test. This separation is an advantage: the team can correct identity or story upstream instead of asking a color pass to solve generation errors.

Choose the hero clip. Generate a small batch from one locked brief, then approve one hero clip. Review it at normal speed, muted, with audio, and frame by frame around motion or lighting transitions. Record why it won and which defects remain acceptable. The decision owner signs the SDR master before Ruby is opened; otherwise two moving targets make the comparison meaningless.

Preserve source quality. Download the highest-quality supported master available from the approved workflow. Do not screen-record it, send it through a messaging app, or re-export it merely to rename it. Preserve aspect ratio, frame rate, duration, audio, and the untouched file. If a delivery platform needs a smaller encode, create that derivative after the HDR decision rather than using it as the test source.

Name and version files. Use names that expose lineage: campaign_shot03_sdr-master_v1.ext, campaign_shot03_ruby-hdr_test01.ext, and campaign_shot03_sdr-delivery_v1.ext. Store the source checksum, Ruby settings, task ID, date, operator, listed estimate, final credit charge, display used, and decision. Never overwrite the SDR master. A new source file requires a new test ID.

Single-render experiment

The test changes one thing: Ruby processing. Keep source, duration, account path, evaluation displays, room lighting, and review order stable. If the tool exposes no adjustable grade controls, record that fact rather than inventing settings.

Test setup. Upload the frozen SDR master through a currently supported Ruby surface. Capture the pre-run estimate and submit one conversion. Download the untouched result. Inspect file duration, dimensions, frame rate, audio presence, and playback before judging aesthetics. A corrupt or mismatched file is a technical failure and should not enter the visual comparison.

Use at least one HDR-capable display with HDR enabled and one ordinary SDR display or SDR interpretation. Label each screen and operating-system setting. Reviewers should see A and B without being told which is expected to win, then record observations independently.

Three-monitor scorecard

Visual QC grid. Review the same frames for highlight detail, shadow separation, skin tones, product color, saturation, gradients, banding, motion stability, text, and unexpected shifts. Require timecodes and screenshots for every rejection. A pass needs intact identity and product facts plus a visible delivery benefit on the target display; “looks more vivid” without a location or comparison is not enough.

Credit estimate. Calculate duration in seconds × 20 listed credits per second for the dated estimate. Then add the actual charge for any permitted retry. Set a one-conversion budget and one diagnostic retry only if a technical failure has a clear cause. Cost per accepted HDR clip equals all Ruby credits spent divided by accepted clips; report the sample size and date, not a universal cost.

Twin-delivery rollback

The final choice belongs to the delivery job, not to the tool. Keep the HDR file only when it passes protected criteria, creates an observable benefit, and has an HDR-capable destination with an approved fallback.

Delivery decision. Publish HDR when the client, player, platform, and display path have been tested with the exact file. Archive it for later when quality passes but the current destination is uncertain. Skip it when the benefit is negligible or the grade creates a protected failure. Record reviewer, date, destination, decision, and evidence link in one row.

SDR fallback. Retain the approved SDR master and a platform-ready SDR derivative. If the HDR upload is rejected, tone-mapped unpredictably, or displayed incorrectly, switch to that known-good file rather than improvising another conversion near deadline. Open the served page or app after upload and compare it with the local master on the intended device.

Archive checklist. Archive the original brief, APOB prompt and inputs, approved SDR master, checksum, Ruby estimate and receipt, HDR output, file inspection, display settings, QC grid, screenshots, reviewer notes, delivery file, and decision. Recheck the Runway model and pricing page by September 20, 2026 or sooner if access, credits, surfaces, or supported inputs change.

One-page decision: approve the SDR source; verify an HDR-capable destination; freeze one file; estimate credits; convert once; inspect the file; compare on HDR and SDR displays; reject any identity, product, motion, text, or color failure; retain an SDR fallback; archive all evidence. The workflow succeeds when it makes the delivery choice explainable—even when the correct answer is to skip HDR.

Sources

• Runway — Product Updates and Changelog

• Runway Academy — Credits and Available Models

• APOB AI — AI Video Generator

• APOB AI — Runway AI Video Generation Workflow

• APOB AI — AI Influencer Generator