Table of Contents Show
Imagen Video applies automated color correction to sequences from Adobe Premiere Pro and DaVinci Resolve, using AI Profiles or uploaded LUTs, then returns a graded project you can adjust in your NLE.
Imagen Video wants to finish the grade before you finish the coffee
Imagen itself describes the tool as adaptive AI colour grading that matches cameras, fixes skin tones, handles exposure and white balance, and delivers a consistent grade across an entire timeline, clip by clip. It supports Adobe Premiere Pro and DaVinci Resolve, and the export/import workflow is built around processed project files rather than a rendered video roundtrip.
The setup is very clear: speed and convenience. This is not a replacement for a supervised grade on a high-end finishing job. The problem being solved here is not “make this commercial look like it came out of a Baselight room with a senior colourist and three clients arguing about teal.” It is more like: match a pile of mixed footage quickly enough that the project can leave the building before everyone involved loses the will to invoice, because they’ll watch it on a non claibrated tablet anyway.
The company comes from high-volume professional photography, with weddings, events, real estate and similar workflows: many files, repeated looks, tight deadlines, limited appetite for manual correction on every shot.
What the tool actually does
Imagen Video works after picture lock. The company recommends uploading a finalised project, because colour correction is applied to the edit as it exists at upload time. The user selects the sequences to process, including nested sequences, and Imagen uploads the project for cloud processing. When the correction is complete, the user receives an email and downloads a new project file.
Corrections can be applied directly to each clip or as adjustment layers. In Premiere Pro, Imagen applies Lumetri Color effects. In DaVinci Resolve, the result is available as node-based adjustments, so the user can still inspect and refine the grade. The processed project file receives “-with-imagen-edit” in its name, while the source project remains unchanged.
The tool supports two colour-direction methods. AI Profiles are pre-made looks built by videographers, which Imagen adapts across the footage. Think Instagram-Filters. Alternatively, users can upload their own LUTs and let Imagen use them as the basis for adaptive colour correction. The company says AI Profiles can be previewed, adjusted and reused, while LUTs can be treated as technical transforms, creative looks, complete LUTs or combinations of technical and creative LUTs.
Imagen’s LUT workflow is the more interesting claim, at least on paper. The company argues that a normal LUT is static: it applies the same transform to every clip, regardless of exposure, white balance, lighting or camera differences. Imagen says it instead uses the LUT as a reference point, then generates clip-specific corrections that preserve the intended look while adapting to the actual shot. For technical LUT workflows, Imagen says users can upload different LUTs for different camera folders. The required LUT format is *.CUBE.
AI Profiles, sliders and the part where taste becomes a settings panel
AI Profiles can be adjusted before colour correction. Users can “Adjust AI Profile,” move sliders to better match their colour style, then save those changes for the current and future projects using that profile. After processing, users can also provide feedback, adjust the profile again, and trigger an updated correction. Imagen then emails the user when the updated correction is ready.
In Premiere Pro, the help article states that the Input LUT is shown as “Imagen Edit” and the adjusted profile changes are visible in the Basic Correction sliders. That is useful because it suggests the result is not a completely opaque rendered grade. Editors can inspect at least part of what was changed and continue working with familiar controls.
Still, this is where the “filter versus workflow” question sits. Imagen insists this is not a generic filter, but a tool that adapts clip by clip. The technical claim is that the grade responds to footage conditions, not that it simply overlays one look on a whole sequence. Fair enough. But from a production standpoint, it still needs to be judged by the output, not the adjective density of the product page.
Where this gets risky
Colour correction across a timeline is not a small automation task. It touches every cut, every lighting shift, every camera transform, every skin tone, every odd white-balance decision and every “we will fix it in post” moment that should have been legally banned around 2009.
Imagen says it can match cameras, balance exposure, handle white balance, preserve skin tones and adapt LUTs across lighting shifts. Those are substantial claims. If the system guesses wrong, the mistake may not be isolated to one clip. It may be spread across a sequence in a way that only becomes obvious when the edit is watched in context – and already downloaded and paid for.
AI systems can produce confident mistakes. They may show up as skin tone drift across a scene, inconsistent contrast between angles, lifted blacks, crushed detail, a strange warmth shift, or a look that feels coherent in thumbnails and wrong in playback. The first proper review should not happen with the client in the room, unless the workflow also includes ritual humiliation as a billable service (Not judging, you do you)
Roundtrip and checking
The nondestructive workflow helps. The original project remains untouched, the graded result comes back as a new project, and the user can adjust clips in Premiere Pro or Resolve. That makes Imagen safer than a black-box render-only system, like a few we decidedly haven’t written about in the last years of AI-everything. But it does not make it safe by default.
The company recommends starting with a shorter sequence before running a full project. That advice is correct and should be taken seriously, despite being buried in product-guide politeness.
Cloud processing, storage and consent
Videos are stored on AWS servers and used exclusively for the user’s projects within Imagen Video. It also says footage is not used for any other purpose without explicit consent and claims GDPR compliance.
For wedding films and corporate work, this may be straightforward, depending on contracts. For agency jobs, broadcast material, unreleased campaigns, sensitive documentary footage or NDA-bound productions, cloud upload is not a small detail.
Pricing
The product page says the first project can be edited for free without a credit card. The pricing page also states that the free trial includes 20 minutes of colour grading.
The listed launch pricing has two individual plans. Pay-as-you-go is billed monthly, with no base fee, and charges only for processed minutes with € 1 per Minute.
The Starter Pack is listed at 30 dollars per month, includes 50 minutes, and charges extra minutes at 0.60 dollars per minute. Unused Starter Pack minutes roll over as credits.
Also, I found no clear information in the available sources about whether minutes are refunded or not charged if the tool fails to apply LUTs as expected. That is exactly the kind of detail that becomes important five minutes after it costs money.
Practical verdict
If you need a first pass for a showreel, a corporate edit, a wedding film, an event recap or a branded sequence with too many cameras and not enough time, Imagen Video could be useful. If you need a final supervised grade for high-end finishing, this is not it, but, to be fair, they do not really claim that either. Which is refreshing, because normally AI tools promise to replace the department, the client, and probably lunch.
https://imagen-ai.com/vblog/meet-imagen-video-professional-color-grading-built-for-editors
