For those who don’t know the tool: Beeble is a AI based VFX toolset built around video to video processing, including SwitchX for background replacement, object inpainting, and relighting, plus Switch Light for generating compositing passes such as normals, albedo, roughness, specular, metalness, depth, and alpha.

The short version: yes, you can use it in Nuke

The Nuke Plugin enables a workflow that stays inside Nuke: load a pass set, connect it to a controller, then drive lighting with dedicated nodes. It targets relighting work where you want quick iterations without building a full CG relight setup from scratch, but still want predictable inputs and knobs you can animate. You need Nuke 14.1v2 or newer. The plugin is working on Windows and macOS. If your show sits on Linux, that missing test line matters.

Install: one folder, one line, one restart

Installation is classic Nuke: unzip the plugin folder, move the Beeble folder into your .nuke directory, then add a single plugin path line to init.py. On macOS, the target path is ~/.nuke/Beeble/. On Windows, it is C:\Users.nuke\Beeble. After restart, the plugin appears on the toolbar as a new menu.

If you want the lowest-drama rollout on a facility machine, treat it like any other Nuke plguin deployment: version the folder, keep init.py changes minimal, and test on a clean user profile before you touch shared configs.

What the plugin actually wants: a very specific PBR stream

The plugin’s key constraint is the PBR multi layer format. It expects a single image stream containing multiple Nuke layers named Source, Normal, Basecolor, Roughness, Specular, and Metallic. The node reference calls this “PBR multi layer stream” the core input, and the controller and light nodes all build on it.

There are three supported ways to feed it: EXR sequences, PNG sequences, or MP4 videos. EXR needs each frame to carry multiple passes packed into layers in a single EXR file per frame, such as Frame_000001.exr. PNG and MP4 need one pass per sequence or video, arranged so the loader can map names to the expected pass inputs.

This is also where the tool draws a hard line between convenience and correctness: if your pass naming or folder layout drifts, import stops being one click.

Load PBR Passes: the most important button is the first one

The Quick Start flow begins in the Nuke toolbar: Beeble, then Load PBR Passes. You point it at an unzipped folder of VFX passes. For PNG, the folder must contain subfolders for each pass, and each subfolder holds a PNG sequence like Normal/Normal_000001.png. For MP4, the folder must include one video per pass named .mp4 such as Normal.mp4.

When you load PNG or MP4, the plugin automatically creates a PBRPacker node to pack the six inputs into a single multi layer stream. For EXR sequences, the PBRPacker is not needed because the passes are already packed into layers. If you ever need to build the stream by hand, the helper menu exposes PBR Packer, and it takes exactly six inputs in this order: Source, Normal, Basecolor, Roughness, Specular, Metallic.

PBR Controller: fix the material before you blame the light

After import, the plugin creates a PBRController node. It exists for one reason: adjust pass intensities and blending before lighting.

Source Blend ranges from 0 to 1 and blends between BaseColor at 0 and Source at 1. Roughness Exposure, Specular Exposure, and Metallic Exposure each run from minus 2 to 2 and adjust intensity in stops. This is the difference between a light that looks wrong and a material that was weighted wrong.

If you plan to publish a facility preset, this is the node to standardize. It is also the node to keyframe if you need a shot that shifts from dry to wet, but still stays within the same pass set.

Lights: three nodes, clear ranges, and a Merge plus

Lighting lives under Beeble, then Lights. The basic rule stays consistent: connect the PBRController output to the light input, tweak properties, then stack more lights with a Merge set to plus. That Merge plus step is explicitly called out for multi light setups.

Directional Light exposes a horizontal rotation from 0 to 360 degrees and a vertical rotation from minus 90 to 90 degrees, plus light color and an intensity multiplier that starts at 0 and defaults to 1.0.

Point Light exposes XY position in pixels, Z depth from minus 5 to 5, light color, intensity starting at 0 with default 1.0, decay starting at 0 with default 2.0 described as inverse square, and maxDistance starting at 0 where 0 means infinite reach.

Environment Light applies image based lighting using an HDRI and takes two inputs: the PBR multi layer stream plus the HDRI. The Quick Start notes you load the HDRI file and connect it to the light input. The node reference describes HDRI as the second input, and the environment light output is RGBA like the other light nodes.

Limits and gotchas you should not ignore

The documentation states the prerequisites clearly: Nuke 14.1v2 or newer, tested on Windows and macOS. It also defines the pass set rigidly as six layers in one packed stream. If your pipeline produces additional passes like depth or alpha, the Nuke Plugin page does not list them as inputs for the PBR stream.

Use it where you can afford to iterate and where the pass set stays stable across the shot. Lock your ingest, standardize your pass naming, then let artists work with Directional, Point, and HDRI lights like they would in any comp, with the added bonus that they can animate light behavior using regular Nuke tools. And yes, test it before you commit. New tools are great at demos and even better at breaking when a client sends a new plate at 2am.

https://docs.beeble.ai/integration/nuke-plugin
https://docs.beeble.ai/tutorial/nuke-tutorial

Suggested tags:

The post Beeble Nuke plugin: relight with passes first appeared on DIGITAL PRODUCTION and was written by Bela Beier.

For those who don’t know the tool: Beeble builds AI relighting and video transformation tools for VFX, post, and virtual production, and Canvas is its very new node based environment that wires SwitchX and SwitchLight into repeatable shot workflows.

A node graph, but for AI outputs you can wrangle

Canvas lands as a node based AI compositing environment aimed at complex workflows across VFX, post, and virtual production. The pitch is straightforward: put AI video models, compositing utilities, and workflow automation into one visual system so teams can iterate across multiple shots without rebuilding the same logic over and over.

In practical terms, it targets the part of the job where you already juggle plates, masks, reference stills, and multiple look directions, then repeat that dance across a sequence.

What is actually in the box

Canvas presents itself as a node graph workflow editor for final render, with a catalog grouped into Input, AI Tools, Edit, and Utility. Input supports text prompts plus image files in PNG, JPEG, or WebP, and video files in MP4 or MOV. In other words, it starts where most editorial and comp workflows already start: media in, decisions next.

https://digitalproduction.com/2026/05/20/beeble-canvas

The AI Tools section includes a node for SwitchX, the same model as the standalone product, with source and alpha handled as video inputs. It also includes an LLM node that can generate files from a prompt with up to four media inputs, plus image generation and video generation nodes that can route to external models. For rotoscoping style work, Canvas includes a Video Matte node that outputs an alpha matte plus RGBA, with an automatic mode and an interactive selection mode, plus options aimed at general footage and green screen keys.

On the classic comp side, the Edit nodes cover the expected utility moves: extract a frame, trim, crop, resize, curves, invert, flip, and morphology operations like erode and dilate. Utility nodes cover housekeeping: sticky notes, routing for cleaner graphs, a viewer, side by side compare, and export. It’s not fully Nuke yet, but…

That combination matters because it lets artists keep the media prep and look development steps next to the model calls, instead of bouncing between separate tools and hoping filenames stay honest.

Relighting and PBR passes, now wired into the graph

The system includes access to SwitchLight for generating physically based rendering passes. In the Canvas node catalog, the VFX Pass Generator extracts PBR material maps including BaseColor, Normal, Depth, Roughness, Specular, and Metallic from image or video using SwitchLight 3.0.

That puts relgihting adjacent tasks into the same graph where you also assemble mattes, feed references, and generate variations. It also matches the stated intent of supporting controllable, iterative pipelines across multiple shots, not one off experiments.

Automation, batches, and the part everyone really wants

Canvas claims to be built around repeatability: build workflows once, apply them to shot sequences, and automate iteration without rebuilding node logic shot by shot. It also calls out batch processing and maintaining scene consistency across projects.

Those claims describe the correct pain point. Anyone who has ever reconnected the same five nodes fifty times will understand why a programmable layer matters. Still, treat the automation story like any new tool claim: test it in your own pipeline with your own edge cases before you bet any delivery on it.

The API angle: SwitchX in your pipeline, with pricing details

SwitchX API for video transformations in YOUR pipeline

Alongside Canvas, the SwitchX API is positioned for studios and developers who want to integrate video transformation and relighting capabilities directly into custom pipelines and software environments.

The developer portal offers a REST style workflow with uploads, job creation, and polling, and it explicitly supports 720p and 1080p output. It also has pay as you go credits in US Dollar, and pricing can start at $0.10 per generation. That pricing applies to the API offering as described there, while Canvas pricing is not specified, since it is part of Bebble Online.

Beeble Studio: 4K AI relighting on YOUR desktop

Availability, media assets, and credits

Canvas is available now via the Beeble platform, with a launch video hosted on YouTube (The Video we embedded above…) .

The boring but necessary reminder

New tools and innovations should be tested before use in production, especially when they touch shot consistency, automation, and delivery formats. Run a small sequence through, break it on purpose, and see how it fails before you trust it on a deadline.

https://docs.beeble.ai/beeble/canvas
https://docs.beeble.ai/beeble/overview

The post Beeble Canvas first appeared on DIGITAL PRODUCTION and was written by Bela Beier.

For those who don’t know the tool: Beeble Background Remover is a Cloud roto and matte extractor for comps, now offering CorridorKey or MatAnyone before upload, so your key tests can happen before your pipeline meeting turns into a sequel.

What changed in Background Remover

Beeble‘s Background Remover now lets you choose between MatAnyone and CorridorKey before you upload footage. MatAnyone stays positioned as the general-purpose option. CorridorKey is positioned as the green screen option, aimed at cleaner edges and more accurate keying on chroma key shots. Background Remover runs only in the Cloud app and does not run in the desktop product.

The fastest way to try it without committing

CorridorKey has the kind of momentum that makes people want to try it immediately, and also the kind of setup that can make them postpone it indefinitely. If you do not want to download code, install dependencies, and explain to your producer why your machine is now compiling something called a model, running CorridorKey inside the Background Remover toolkit gives you a low-friction way to test it on your own footage first. That includes testing it against your current keys, and checking whether your exports land cleanly in your comp toolchain, without rebuilding that toolchain just to run an experiment.

How CorridorKey was invented

CorridorKey started as a response to a workload problem. A greenscreen-heavy show needed a lot of shots, and the team wanted an automated way to cut subjects out so they could spend time on the creative side of VFX instead of hand-cleaning keys frame after frame.

The design focus landed on a specific compositing headache: a greenscreen edge pixel often contains a mix of foreground and background light. A matte can be good and the color can still be wrong, because the plate already baked the screen contribution into semi-transparent detail. The CorridorKey approach targets a straight color foreground output paired with an alpha output, so semi-transparent materials, motion blur, and hair can keep behaving like actual transparency instead of looking like tinted cutouts.

Training data was coming from CG so the model could learn from mathematically clean answers. The workflow renders greenscreen plates, then renders the same foreground without the background and with a perfect alpha, giving the model ground truth for both color and transparency. Scale was coming from procedural variation, including procedural scene generation work in tools like Houdini and Blender to produce many variations in lighting, materials, objects, and motion without rebuilding each shot by hand.

Early results were treated as tests and iterated on through retraining. One tactic for pushing quality was to composite results over random backgrounds and neutral gray during testing, so leftover contamination becomes obvious and gets punished during training. The tool is being used as a first step in a broader pipeline, with outputs intended to drop into comp and realtime contexts in apps like Adobe After Effects and Unreal Engine.

What you get out of Beeble after a run

If you are testing CorridorKey for greenscreen work, this matters because the outputs determine how painful your handoff is. A quick trial that ends in a format conversion detour still counts as pain. It just wears a different hat.

Background Remover is built around alpha extraction for compositing and VFX workflows. You can upload a video, an image sequence, or a single image, then download results intended to slot into a comp.

On the Cloud pricing page, export options are listed as RGBA exports in MOV or PNG sequence and alpha exports in MP4 or PNG sequence. MOV here refers to the QuickTime container, PNG refers to the PNG format, and MP4 refers to the MPEG family container.

Pricing, credits, and practical limits

Background Remover usage is credit-based. The Cloud pricing is around image uploads as consuming one credit per image and video uploads as consuming one credit per second of video.

The plans listed include a free Starter tier that includes ninety credits per month and is labeled non-commercial. Two paid tiers have Background Remover included. “Creator” is listed at nineteen US dollars per month or sixteen US dollars per month when billed annually, with five hundred forty credits per month. “Professional” is listed at seventy-five US dollars per month or sixty US dollars per month when billed annually, with two thousand four hundred credits per month.

Per-upload maximums for video length are thirty seconds on the five hundred forty credit tier and one minute on the two thousand four hundred credit tier. There is also a batch upload up to twenty files at once on the paid tiers.

A production reality check

CorridorKey is a model option, not a replacement for judgement. Every keying tool behaves differently across compression, sensor noise, spill, lens artifacts, and the very special chaos of talent wardrobe choices. Test CorridorKey on your own porudction plates before you trust it with a deadline, and keep your existing approach ready for the shots that refuse to behave.

If the results look good, treat the first integration step as a controlled experiment: run a small batch, validate edge stability, validate color behavior in the semi-transparent areas, validate the exports in your comp, then scale up. That is less romantic than a revolution, but it is how tools earn their place in compositnig pipelines.

https://docs.beeble.ai/beeble/background-remover

https://github.com/nikopueringer/CorridorKey

https://beeble.ai

The post Beeble adds CorridorKey green screen model first appeared on DIGITAL PRODUCTION and was written by Bela Beier.

For those who do not know the tool: SwitchX is Beeble’s image and video transformation system for prompt-driven and reference-driven visual changes such as lighting adjustment, background replacement and related compositing-style alterations. Until now, that workflow lived primarily in the web application.

Beeble has launched the SwitchX API in public beta, taking its browser-based image and video transformation tool and turning it into something developers and pipeline teams can actually integrate into production systems. That is the useful part of this release. The familiar launch language about cinematic transformation is there, of course, but the more relevant shift is structural: SwitchX is now available via a documented REST API rather than only through a web interface.

From web app to pipeline component

SwitchX API follows a fairly standard service model. Users authenticate via x-api-key, upload source media, start a generation job, then either poll the result endpoint or use a callback URL for webhook-based completion handling. In other words, it behaves like an API, which should not be remarkable, but in this market it sometimes still is.

That matters because it moves SwitchX from a browser-first tool toward something that can sit inside internal apps, automated workflows, or technical artist tooling. For teams trying to reduce manual handling between editorial, compositing, review and delivery, that is a more meaningful step than the launch itself.

What the API actually supports

The generation endpoint accepts both image and video jobs. Source material can be supplied from Beeble-hosted upload URLs, external HTTPS URLs, or inline base64 payloads up to 50 MB. Supported image formats include PNG, JPEG and WebP. Supported video formats include MP4 and MOV using H.264 or HEVC, with a maximum of 240 frames per job.

Output is (currently) limited to 720p and 1080p, with 1080p being the default. That does not place SwitchX in final-finishing territory, but it does make it usable for iteration, look development, internal tool integration, and some fairly direct production use where speed and automation matter more than absolute output scale.

Alpha handling

SwitchX supports four alpha modes: auto, fill, custom and select. Auto handles subject separation automatically. Fill keeps the full frame. Custom accepts a supplied alpha asset. Select allows users to define a first-frame alpha keyframe that the system then propagates across the shot.

Background replacement, relighting and subject isolation all sound smooth until edges, motion blur, hair and semi-transparent detail arrive and start asking difficult questions. The API exposes enough control to make the workflow more testable and less opaque.

Webhooks, retries and other unglamorous useful things

Beeble has also included a few details that make the SwitchX API more realistic for technical workflows. The start-generation endpoint supports callback_url for webhook-based completion and idempotency_key for safer retries. This suggests Beeble is not only targeting solo browser users but also teams that want to integrate SwitchX into repeatable systems. That does not automatically make it production-proven, but it does mean the API is designed with awareness of how production systems tend to break.

Pricing is prepaid and usage-based

Beeble is positioning the SwitchX API as a pay-as-you-go service with prepaid USD credits. The public pricing page lists Build-tier pricing at $0.10 per 30 frames for 720p video and $0.30 per 30 frames for 1080p video. Images are priced at the same two levels. The minimum top-up is listed at $50, and the default monthly spending cap is $5,000. At launch, the API appears to focus specifically on SwitchX. Other Beeble tools, including SwitchLight, are not currently presented as API-accessible products in the same way. We’ll see what happens :)

What this means in practice

The interesting part of the SwitchX API launch is not the promise of cinematic transformation without manual VFX work. The interesting part is that Beeble now offers a well-documented service with uploads, generation jobs, polling, callbacks, account controls and usage-based billing. That moves SwitchX from a browser feature to a pipeline candidate.

Whether it becomes a serious production tool will depend on output consistency, edge quality, latency, and how much cleanup still lands back on artists after the automated step has finished. That question remains open, which is normal. The difference now is that developers and technical teams can test it directly inside their own workflows instead of evaluating it from the sidelines through a browser demo.

https://developer.beeble.ai/

The post SwitchX API for video transformations in YOUR pipeline first appeared on DIGITAL PRODUCTION and was written by Bela Beier.

For those who don’t know the tool: Beeble runs AI relighting and pass generation in the browser via the Cloud app and locally via Beeble Studio, using SwitchLight for video-to-PBR work and SwitchX for background swaps, with exports that slot into comps in Nuke and motion work in After Effects.

Background Remover does not add a brand new trick. It pulls the same AI masking model used in VFX Pass Generator into a standalone, faster workflow meant for the quick matte job, with the provided materials calling it more credit-efficient than a full pass as a marketing claim. Exports lean hard into practical comp handoffs, including RGBA video with alpha, RGBA PNG sequences, and alpha-only sequences, which fits motion graphics work where you want clean holdouts so titles and design layers can live behind moving subjects.

For more background, see Digital Production coverage of SwitchX background swaps, Studio local 4K relighting, and SwitchLight 3.0.

The real news is the shortcut

Background Remover lives inside the Cloud app as a standalone alpha extraction workflow. The core promise is not mystery science. It is a packaging change that targets one task, foreground mattes, with fewer steps and fewer credit demands than running a broader multi-pass job.

That matters because a lot of shots do not need a full buffet of passes. Plenty of days you just want a clean subject cutout so you can comp a replacement plate, push a quick relight downstream, or build motion graphics behind a moving performer. Background Remover aims at that moment when you would normally sigh, open your roto tool of choice, and start paying for coffee by the hour.

It also draws a bright line between where it runs and where it does not. Background Remover is available exclusively in the Cloud app and it is not available in the desktop product yet.

Inputs that match real-world chaos

The tool accepts three kinds of input: a single image, a video, or an image sequence. Single images support common formats, including PNG and JPEG. Image sequences support up to 2,000 frames. For video, the supported container and codec combinations include MP4 using H.264 and MOV using ProRes 422, ProRes 4444, or H.265. The maximum file size is 2 GB. The tool supports up to 2K resolution at 2048 pixels, and it will downscale larger video. It supports clips up to 60 seconds and up to 30 fps, with higher frame rates reduced.

Color handling has a practical bias. Rec. 709 for video and sRGB for images are called out as the best quality settings, while other color spaces such as Log can degrade quality. That is less of a moral stance and more of a reminder that alpha extraction likes predictable input, not surprise curves.

Preview first, then commit, unless you like living dangerously

Background Remover runs a two-phase workflow for video. First comes a preview stage that extracts a preview frame and generates rough alpha previews. This stage exists for sanity checking. The preview alpha is explicitly rough, not pixel-perfect, and fine details like hair edges and semi-transparent areas are refined during full extraction.

If you want to steer the result, there is an interactive refinement step. The workflow uses an Edit Alpha action inside the preview dialog. That opens a rotoscoping UI designed around quick subject selection. The demo workflow shows clearing the existing mask and selecting only the subjects you want to keep by clicking them, then applying the alpha.

There is also a Skip Review mode. It bypasses preview and refinement and runs full extraction immediately. In that mode, the first frame becomes the reference and the pipeline runs without manual review. It is intended for simpler shots with clear subjects and consistent backgrounds. None of this replaces judgment. It simply gives you the option to front-load a small amount of control before you burn the full render.

Exports that behave in comps

Background Remover produces two output types: RGBA results with transparency applied, and grayscale alpha mattes. For RGBA, it can output a MOV file with an embedded alpha channel, plus RGBA PNG output as a sequence or a single image. The MOV output is 10-bit RGBA in sRGB. The PNG RGBA output is 8-bit in sRGB.

For alpha-only delivery, it can output an MP4 grayscale matte as well as grayscale PNG output as a sequence or a single image. Grayscale alpha outputs are 8-bit and represent the matte range from 0.0 to 1.0.

The point here is pipeline friction. An embedded-alpha MOV can drop straight into After Effects, DaVinci Resolve, or Final Cut Pro without juggling a separate matte file. When you do want a separate matte, the alpha-only options cover both video-centric and frame-centric workflows.

If you build motion graphics, this is the part that keeps paying rent. A clean matte lets typography, UI, and graphic elements sit behind a moving subject while staying readable. That trick stays stubbornly useful, even when the rest of the shot changes.

Credits, pricing, and the part everyone asks first

Pricing is expressed in credits and it is published on the official pricing page for the Cloud plans. Background Remover is listed at 1 credit per second of video, and 1 credit per image. VFX Pass Generator is listed at 3 credits per second of video, and 3 credits per image. That is the math behind the one-third claim.

Plan details and credit bundles live on the Cloud pricing page, including per-plan limits. Note that Background remover is not included in the free tier.

Limits that matter when you are on a deadline

Image and image-sequence uploads have a maximum of 100 MB per file. Resolution supports up to 2K at 2048 pixels, and larger files are rejected. Video uploads support up to 2K, with larger video downscaled. Video duration supports up to 60 seconds and frame rates up to 30 fps, with higher rates reduced.

Bulk handling exists too. The tool supports bulk downloading multiple completed results at once. For anybody juggling variants, alt takes, or multiple elements, that avoids the classic cloud workflow of clicking download until your wrist files a complaint.

Where it fits in the larger toolset

Background Remover sits as a focused subset next to the broader pass workflows. If you only need a matte, it keeps you out of the heavier multi-pass path. If you need more, the pass generator remains the place for depth, normals, and other map outputs that support relighting and downstream work in 2.5D and 3D contexts.

https://app.beeble.ai/

The post Beeble Background Remover hits Cloud app first appeared on DIGITAL PRODUCTION and was written by Bela Beier.

For those who don’t know the tool: Beeble is a cloud-based VFX relighting platform. Its new SwitchX adds generative background and lighting changes inside the same web app, not as a DCC plug-in.

Backgrounds first, everything else optional

Beeble has introduced SwitchX, a generative video feature designed to make background and lighting changes straightforward while keeping the main subject intact. SwitchX allows artists to replace environments, adjust lighting, and modify elements such as props or costumes while preserving the recognisable appearance of the subject. The core idea is simple: switch the world around the subject without altering what matters. The tool runs entirely inside the Beeble cloud application. No desktop version or direct host integration is mentioned in the official material.

A simple cloud workflow

The workflow centres on ease of use. Users upload a video or image to the Beeble web app. SwitchX automatically detects the main subject, and an interactive masking tool is available for finer control. A reference image is then added to define the new background, lighting and mood. The system analyses the footage frame by frame and pixel by pixel to guide the generation process and maintain temporal consistency.

In a few minutes, SwitchX can generate a 2K video with a new background, updated lighting, shadows and adjusted camera perspective. Identity, facial characteristics and audio sync are intended to remain intact. The exact pixel dimensions, codec and colour pipeline are not specified.

Relighting and environment reshaping

Beyond swapping backgrounds, SwitchX enables relighting of entire scenes. Mood, time of day or genre can be shifted by adjusting lighting conditions through the reference image. Props, costumes and set elements can also be modified. The material describes physically correct light interaction, though no technical definition or rendering model is detailed.

The launch video on YouTube demonstrates examples such as turning an everyday room into a studio-style setting or placing a subject into a city street environment. These examples illustrate aesthetic possibilities rather than measured performance data. Camera perspective adjustments are mentioned, but it is not clarified whether these involve 3D reconstruction or 2D approximations. There is no information about depth passes, AOV export or multilayer output.

Output and beta access

SwitchX produces 2K output in the cloud. The term crisp is used descriptively, without technical metrics. Bit depth, colour space, HDR support and ACES compatibility are not specified. The transcript of the launch video notes that SwitchX is available for free during a beta period. Pricing beyond beta is not included in the official announcement. There is no detail on render limits, clip length restrictions or commercial licensing terms.

Data handling, encryption and storage policies are not addressed in the provided material. Studios working with sensitive footage will need to confirm these aspects directly with Beeble.

Practical take

SwitchX focuses on easy background switching combined with relighting, delivered as a cloud-based video-to-video service. It promises consistent 2K output in minutes and aims to minimise flicker and drift. Whether it meets production tolerances will depend on shot complexity and delivery standards. As with any AI-driven tool entering a post pipeline, thorough testing is essential before committing to client work.

https://beeble.ai/research/switchx

The post Beeble SwitchX focuses on easy background swaps first appeared on DIGITAL PRODUCTION and was written by Bela Beier.

Beeble Studio is the desktop version of Beeble, designed to run entirely on a user’s workstation. It uses the company’s latest AI model, SwitchLight 3, to convert ordinary video footage into a full suite of VFX passes: normals, base-color, metallic, roughness, specular, alpha mattes, and depth maps. The goal is to enable a “video to VFX” workflow: give it footage, green-screen or not, and get back “2.5D assets” that can be relit, composited, and integrated into 3D environments or compositing pipelines with realistic light behaviour.

Technical Features and Pipeline Details

Beeble Studio accepts native 4K footage and can process video clips up to one hour in length in a single render job. The processing happens fully on the user’s machine using the local GPU, with no upload or cloud involvement. Rendered output can be exported as multi-channel 16-bit EXR files containing all AOV passes (normals, base-color, metallic, roughness, specular, alpha matte, and depth) or as PNG or MP4 if desired.

The internal Beeble Editor allows artists to relight footage in real time. Users can place point lights, area lights, environment HDRIs, or simulated video lights, animate lights over time, adjust depth and material responses, refine alpha masks, and blend 2D and 3D elements. A 3D viewport is available for physically correct camera and lighting control, alongside a 2D preview for fine image-level tweaks. Integration plugins are provided for popular DCC and compositing applications such as Blender, Unreal Engine, and Nuke, making it possible to bring the generated PBR passes into standard production pipelines.

Input formats supported include EXR, PNG, MOV (for example, ProRes), or MP4. Export formats include EXR, PNG, or MP4. Batch processing is supported via a render queue. Advanced controls such as “deflicker” and toggles for render parameters are available to help manage temporal consistency and output quality.

Beeble’s technical specification for VFX-pass output defines normal, base-color, metallic, roughness, specular, and alpha as standard 8-bit maps (PNG or MP4), with depth map exported as a 16-bit float EXR for better precision. Normal maps are in object space (OpenGL format). The depth map is a normalized grayscale within the foreground range. This setup follows typical PBR workflows where scene geometry and material properties are inferred from 2D footage rather than full 3D scans.

Improvements in SwitchLight 3

The core change in version 3 of the model is that it treats video as true video. Instead of processing each frame in isolation and then deflickering after rendering, SwitchLight 3 processes multiple frames simultaneously. This delivers more stable temporal results with less ghosting, fewer flicker artifacts, and better continuity in motion sequences. The training dataset for SwitchLight 3 is reported to be ten times larger than the one used for the prior version. This broader data foundation helps the model better understand a wider variety of lighting conditions, materials, and scene types, which should translate into improved realism and generalisation across differing shots.

In practice, Beeble claims that surface details such as skin texture, hair strands, fabric weave, or small geometric cues survive relighting better than before, and that artifacts present in older versions such as odd normal-map tinting or flattening under shadows are substantially reduced. SwitchLight 3 also improves edge consistency and temporal coherence between frames. This reduces shimmer and incorrect shadow transitions when playing back moving footage, which was a common limitation in earlier models.

Use Cases, Practical Considerations and Limitations

Beeble Studio is targeted at VFX compositors, lighting and look-dev artists, small to mid-sized studios, independent filmmakers, and virtual production workflows that lack access to full 3D scans or elaborate set lighting. The tool offers a route to relight live footage, perform background replacements, or integrate real footage into CG scenes with minimal setup compared with traditional photogrammetry or scanning-based pipelines.

However, the quality of AI-generated passes is strongly dependent on the source footage. For best results, footage should feature clear subject separation, sharp focus, high bitrate, correct colour space (Rec.709 or sRGB for video or PNG/JPG for images), good exposure and lighting, and minimal extreme camera motion or motion blur. Compressed, noisy or low-detail footage, or video shot in non-standard color spaces, may degrade final PBR output.

If footage includes a proper alpha (for example, ProRes 4444), Beeble Studio can use that directly instead of attempting AI-based background removal. Otherwise it will attempt to generate a clean foreground matte.

Users adopting Beeble Studio should evaluate its output on a case-by-case basis. While SwitchLight 3 offers improved fidelity, AI-driven relighting currently remains a complement, not a replacement, for dedicated photography, lighting setups, or traditional 3D-scan workflows.

At press time, these technical specifications and claims are drawn from Beeble’s published documentation and vendor announcements. Studios and post-production teams planning to integrate Beeble Studio should run their own tests with real-world footage to confirm whether output meets their pipeline standards.

The post Beeble Studio: 4K AI relighting on YOUR desktop first appeared on DIGITAL PRODUCTION and was written by Bela Beier.

Beeble has released Beeble Camera, a free iOS app that uses Apple’s LiDAR and ARKit frameworks to track iPhone movement in real time while recording. The app anchors captured footage to real-world space and uploads it directly to the Beeble Web platform for AI processing. According to the developer, uploaded shots are processed into camera-tracked, rotoscoped and relightable VFX passes. These passes can then be composited in Beeble Studio, the online Beeble Editor, or exported into Blender and Unreal or using a dedicated plugin, or any other DCC through a straight export.

Setup and Shooting

After installing Beeble Camera from the App Store, users must scan their environment with the iPhone to allow ARKit to detect surfaces. A floor grid can be tapped to set the origin point for tracking. The app also allows insertion of reference geometry by placing a .usdz file into the scene. This can be scaled, rotated and moved as needed. Once the environment is prepared, recordings capture both video and camera metadata. For improved results, Beeble recommends using a gimbal, ensuring a clean lens (D’uh), and shooting under even lighting conditions.

Uploading and Processing

Recorded clips are uploaded from the phone to Beeble Web, where AI routines create roto masks and relightable passes. These are automatically aligned with tracked 3D data from the shoot. The same 3D reference model used on-set can be converted to .glb and imported into the Beeble Editor, where metadata ensures accurate placement in the composite. Alternatively, Blender users can bring in the footage via the Beeble add-on, which carries over the camera-tracking data.

Availability

Beeble Camera is available now as a free download on the iOS App Store. Processing and compositing features require a Beeble Web account.

What’s Next from Beeble

Beeble CEO Hoon Kim has confirmed plans for a local version of the platform in future updates. He discusses this roadmap and accessibility of AI-based relighting in our Digital Production interview, so read on over there :)

We have to talk about Switchlight 2.0

The post iPhone for Tracking: Beeble Camera enables Free “Virtual Production” first appeared on DIGITAL PRODUCTION and was written by Bela Beier.

DP: Let’s kick things off with a bit of background: What actually set you and your crew on the path to AI relighting? 

Hoon Kim: Four other technologists and I founded Beeble as a selfie relighting app after seeing a gap in the market for AI-powered lighting tools. It received a great reception and went viral, but the real Eureka moment happened next. We found that VFX artists would hugely benefit from a similar tool that relights video instead of still imagery. We knew it would solve many time and budget issues, while enabling professionals to fix lighting in post, without having to reshoot. 

DP: With Switchlight 2.0 being released versus 1.0: What’s the “wow, that’s new!” factor? 

Hoon Kim: Switchlight 1.0 already had a huge level of detail in terms of PBR map generation, but this update brings it up a notch. It boosts the quality of our normal, basecolor, metallic, roughness, and specular passes, thanks to training on 13 times more data. We’ve now also extended these maps to background objects and environments, as well as the central video subject. 

DP: For anyone curious to jump in: How does the pricing and licensing for Switchlight 2.0 work?

Hoon Kim: All our pricing is credit-based, giving you a set amount of content to process each month. The Starter plan is free and allows you to run 15-second video passes, ideal for testing the software. For ongoing use, the Creator plan is $19/month and the Professional plan is $75/month, with discounted rates available on annual subscriptions. All pricing details are available on our site.

DP: If you think about Switchlight landing on the desk of a busy VFX artist or colorist: Where does it shine? 

Hoon Kim: Put simply, it delivers the level of quality a VFX artist would expect. Additionally, they receive AOV passes almost instantly, a significant shift from the traditional, manual CG rendering process, making it ideal for quick fixes.

DP: Deflickering temporal noise is a game-changer for a lot of us! Can you give us a peek into how your new approach works, especially for fast-paced or multi-source lighting?

Hoon Kim: SwitchLight 2.0 operates on an image-based model, utilising PBR maps that mitigate temporal noise deflickering. Our next update, SwitchLight 3.0, takes things further: a native video model that removes the need for special deflickering. More details coming soon!

DP: Speaking of standards: What’s your support like for things like ACES, Rec. 709, P3, and other pro color spaces? Are you outputting true 16-bit or 32-bit float EXRs, so no one’s losing data along the way?

Hoon Kim: Our AI is trained on the Rec. 709 colour space, which is the recommended format for best results. That said, you can still apply the generated PBR maps to your original ACES footage without losing any detail.

DP: What kind of shots or material have you seen Switchlight absolutely nail and where does it still sweat? 

Hoon Kim: SwitchLight 2.0 is built to handle footage from any camera, whatever the quality, lighting, or style. Of course, high-definition, steady shots with flat lighting will always give you the greatest control.

DP: When you’ve got those nightmare shots with crushed blacks, low-res and ancient TV rips: Do you have any hacks? 

Hoon Kim: For that kind of challenging footage, it helps to upscale before running it through Beeble, and then apply a separate deflicker pass on top of the generated PBR maps. Tools like Nuke Copycat make that step straightforward and can dramatically improve the result.

DP: Any “dos and don’ts” from what you’ve seen in the wild?

Hoon Kim: AI upscaling is definitely a “do”, it could massively improve poor-quality footage and makes it easier for SwitchLight to generate accurate PBR maps. But avoid heavy denoising, which strips away detail the AI needs, and hold off on strong LUTs or oversharpening until after relighting, as they can distort the data and lead to unstable results.

DP: Is there a “Switchlight for Dummies” checklist for prepping shots? 

Hoon Kim: We actually put together an article on this! However, my main advice is simple: aim for even, level lighting when shooting. Beeble can handle almost any condition, but flat lighting gives you the most flexibility and control later in post.

DP: Alright, time for a technical deep dive! How does Switchlight 2.0 whip up those depth maps? 

Hoon Kim: At its core, it’s a neural net doing the heavy lifting. That’s about as far as I can go! You can more information via these links: https://beeble.ai/research/switchlight-2-0-is-here or the paper on the technology: SwitchLight: Co-design of Physics-driven Architecture and Pre-training Framework for Human Portrait Relighting

DP: Under the hood, what kind of AI are we talking about? 

Hoon Kim: We are always improving the network architecture, so “monster hybrid” would probably be the most accurate description. However, the beauty lies in integrating the Cook-Torrance BRDF model, which enables the AI to learn how light bounces off a subject. Paired with physically based rendering, the system can generate a relit scene that looks physically convincing.

DP: The classic pain point: specularity. How does Switchlight tease out reflections and highlights? 

Hoon Kim: We’ve made specular maps flexible enough that you can dial them in however you like. Want more gloss, less sweat? Just tweak the strength directly in-platform, or plug the maps into a third-party shader like Blender or Unreal Engine for full control. You can even isolate high- or low-specularity regions, so you decide exactly how shiny or matte things should look.

DP: What about your reference data? 

Hoon Kim: All of our training data is created in-house using scans of real humans and objects. We train only on safe, clean, and fully licensed datasets, ensuring professionals can use our service with complete peace of mind. We do not use movies, films, or any third-party content that could raise copyright concerns..

DP: Platform time! So you’re solid on Windows, Red Hat, Ubuntu. Any dreams of a Mac version for the “Final Cut Victims”? 

Hoon Kim: Right now, SwitchLight 2.0 and the Beeble platform are web-based. With 3.0, we plan to introduce a local version, one that better aligns with the security and workflow standards of major post-production studios.

DP: For anyone building a new Switchlight workstation: What’s the minimum gear to get real-time, and what’s your dream setup? 

Hoon Kim: We’ve designed Beeble to be accessible to anyone, whether you’re a content creator on a laptop with phone footage or a studio running complex plates. At the entry level, all you really need is a laptop and some footage to upload; the platform handles the heavy lifting and generates the VFX passes for you. Bigger studios will naturally push higher-res material through, but it’s not a requirement. And for anyone wanting to get started right away, we’ve just released Beeble Camera: a free app that brings camera tracking to Beeble.

DP: Looking into the crystal ball: Where do you see Switchlight and AI relighting going, say, five or ten versions from now?

Hoon Kim: Looking ahead, our goal is to keep pushing the detail and fidelity of the PBR maps while advancing motion capture, camera tracking, and depth generation. Ultimately, we want to break footage down into its most detailed components so every frame can be reconstructed, relit, and reimagined with complete creative freedom.

DP: For the tech-heads and code nerds: Are there research papers or blog posts you’d absolutely recommend to get your head around the science powering Switchlight? 

Hoon Kim: Absolutely. Our paper is here for anyone to read. It provides a detailed overview of the methods behind SwitchLight, from self-supervised training to the physics-based models that underpin our approach. 

DP: What’s the most fun or jaw-dropping thing you’ve seen Switchlight do that made you stop and say: “Okay, that’s actually impossible—except now it isn’t”?

Hoon Kim: A standout for me was a father-and-son duo who made a fan film called Star Wars: Entrenched, with Beeble helping along the way. Navaz Dowling, who runs Bad Beetle Entertainment, taught his son the basics of VFX and brought in friends for costume design and relighting. The trailer’s already out, and the post-production community has responded really positively.

At the same time, we’re seeing larger post-production studios integrate Beeble into their pipelines. The work Boxel Studio did on Superman & Lois is a great example of this. With SwitchLight 3.0 introducing a local version, that side of adoption is only going to grow.

The post We have to talk about Switchlight 2.0 first appeared on DIGITAL PRODUCTION and was written by Bela Beier.

Beeble’s SwitchLight 2.0 marks the company’s push to put AI-powered relighting into the hands of actual postproduction artists—and not just demo video influencers. The new release introduces physically based rendering (PBR) relighting, portrait and universal HDRI extraction, AI-based deflicker, a neural detail enhancer, and local processing for the security-conscious (or just the deeply skeptical). All these features target one goal: letting post artists relight filmed footage with actual, production-grade material and lighting data, without needing a full 3D scan or reshoot.

From Green Screen to Full PBR Passes—What’s New in 2.0?

SwitchLight 2.0 can generate a complete suite of PBR passes from video footage—including normals, depth, albedo, roughness, specular, and ambient occlusion. These can be exported as layered EXRs or video, suitable for use in Blender, Unreal Engine, or compositing apps like Nuke and After Effects. The updated system also brings improved halo handling, temporal stability, and smart anti-ghosting—so even when the footage is a bit rough, Beeplle claims to keep edges tight and details intact.

HDRI Extraction for Real-World Lighting

A major new feature is dual HDRI extraction. “Portrait HDRI” mode analyzes the lighting on a subject’s face, then generates an HDRI map of the environment—useful for relighting close-ups or matching CG light to live-action. “Universal HDRI” uses Beeble’s DiffusionLight technology to estimate lighting for entire scenes, not just heads. Both methods aim to shortcut the process of shooting reference balls or fiddling with 360-degree cameras (switchlight-studio.beeble.ai).

Neural Enhancers and Deflicker—But No Magic

Beeble incorporates AI-based deflicker to tame temporal noise, and a neural enhancer that tries to restore details like subsurface scattering and self-occlusion lost to compression or lighting limitations. That sounds like a marketing claim, but the neural enhancer’s job is to create passes that don’t look like synthetic plastic when relit. In practice, expect results to vary based on source quality. Still, these tools are designed to reduce the time artists spend cleaning up messy mattes and reconstructing details by hand.

The Workflow: From Upload to Export

The workflow starts in the Beeble Studio web app or desktop version: upload video (green screen or not), and the AI gets to work—extracting backgrounds, creating PBR passes, and providing a real-time interface for relighting. Within the Studio, users can swap HDRIs (from the built-in library or custom sources), add point lights, and preview the results with instant feedback.

Need a background? Text prompts can generate new AI backdrops on the fly, with the usual caveats of GenAi. When finished, export your full composite or individual passes in 4K+ for Blender, Unreal, Nuke, or straight to edit. A dedicated loader plugin lets Blender and Unreal Engine users import PBR assets with a click, skipping the usual roundtrip of manual pass matching.

Deployment: Web, Desktop, CLI—And Air-Gapped

Beeble offers both web and desktop versions. Local processing is the default; no uploads to the cloud unless you want them. Air-gapped studios get RLM licensing and command-line (CLI) batch processing on Windows or Linux—making it possible to integrate into automated pipelines or run batch jobs on render farms without an internet connection (switchlight-studio.beeble.ai).

Who Should Use SwitchLight 2.0?

SwitchLight is not for Instagram filter enthusiasts. Its core audience: VFX compositors, lighting artists, colorists, look-development supervisors, indie filmmakers, and virtual production teams who need to relight live-action or green-screen plates with credible, production-ready PBR data. It’s also a fit for small shops and individuals who want the flexibility of PBR relighting without a massive pipeline investment or multi-camera setup. In practical terms, anyone who wants to blend real and CG elements (or relight plates to match different CG environments) without access to a full 3D scan should consider trying it.

Caveats: No AI Is Infallible

AI’s strengths—speed and automation—don’t make SwitchLight immune to artifacts, lighting mismatches, or temporal hiccups. The more you push the relighting or try to extract passes from low-quality or compressed sources, the more likely you are to see oddities. Beeble CEO Hoon Kim specifically urges professionals to validate AI-generated passes and composites in their own postproduction pipelines before trusting the results for final delivery (vp-land.com). That’s not just sensible—it’s necessary for any artist who wants to keep their QC department onside.

Reminder:
Innovative relighting and AI pass generation tools like SwitchLight 2.0 are best tested thoroughly in your own pipeline before any real-world delivery. No exceptions.

beeble.ai

The post Beeble: AI relighting steps up to validate in your pipeline first appeared on DIGITAL PRODUCTION and was written by Bela Beier.