Articles

Advertising VFX Workflow: From Plate and Simulation to Final Composite

ARTILABZ™

ARTILABZ™ gives you unlimited access to all Houdini courses, 3D assets, simulation files, textures and tools. updated every month.

Everything You Need to master Houdini.

01

Premium Houdini Tutorials

Full access to every course — fluid simulation, procedural FX, brand visuals and more.

02

Monthly New Content

Fresh tutorials and assets added every month — your library grows with you.

03

Instant Access to Everything

The moment you join, the full library is yours — no drip-feed, no waiting.

04

Project Files Included

Every tutorial comes with the full Houdini scene file — open every node, learn every detail.

FROM 14.99€/MONTH

Includes one exclusive complete course

The exclusive course — a full production tutorial you won't find anywhere else, never sold alone.

Best Seller
Most Loved
Tutorial Camera Rig

ADVANCED CUSTOM CAMERA RIG

ANIMATION · CONSTRAINTS · CUSTOM UI

BUILD A FULLY CUSTOM CONSTRAINT-BASED CAMERA RIG IN HOUDINI WITH A CUSTOM UI PANEL. DESIGN FLEXIBLE SYSTEMS FOR PRECISE, CINEMATIC CAMERA ANIMATION ON ANY PROJECT.

€29.99

Freebies
Free Studio HDRI Pack box by Artivoxa showing 60 studio lighting setups with softboxes wrapped around the packaging

Studio HDRI Collection

ASSETS · EXR & HDR · 60 HDRIS

DOWNLOAD 60 STUDIO HDRIS CAPTURED IN A REAL PHOTO STUDIO. LIGHT YOUR PRODUCT AND BEAUTY RENDERS LIKE A PHOTOGRAPHER — SOFTBOX, LANTERN, STRIP AND GRID SETUPS, READY FOR ANY RENDERER.

FREE

Advertising VFX Workflow: From Plate and Simulation to Final Composite

A professional advertising VFX workflow turns a creative brief into a technically defined shot, then carries that definition through plate preparation, matchmove, asset work, Houdini simulation, rendering, compositing and delivery. Each stage depends on decisions made upstream: an inaccurate camera solve undermines integration, incorrect scale destabilizes simulations, and incomplete render passes restrict the final composite.

This guide maps the complete commercial VFX pipeline for plate-based, full-CG and hybrid advertising shots. It explains the technical choices and handoffs at each stage, including lens distortion, color management, simulation caching, AOV planning and Nuke compositing. It also identifies the failure modes that most often make CG appear detached from a plate—and the review checks that catch them before final delivery.

1. Define the shot before production begins

A successful workflow begins with a shot definition, not a modeling or simulation task. Before artists build assets or create Houdini networks, the team should agree on what the audience must see, which elements will be digital and what the final delivery requires. Those decisions determine asset detail, simulation resolution, render strategy, review milestones and compositing scope.

Choose between a plate-based, full-CG or hybrid shot

A plate-based shot starts with live-action footage that establishes the camera movement, environment, lighting cues and often practical interaction. CG may replace or enhance a product, add effects, or create shadows and reflections that were not captured on set. The real-world context can support convincing results, but it makes camera tracking, lens matching and plate preparation critical.

A full-CG shot creates the camera, product, environment, lighting and effects entirely in 3D. This approach suits impossible environments, tightly controlled product visuals and campaigns that need many views or formats from the same assets. Its main trade-off is that the team must create every cue that live action would normally supply, including scale, surface variation, environmental reflections, motion blur and depth of field.

A hybrid shot combines both methods. A filmed background might remain intact while a CG product, simulated liquid and digitally controlled reflections occupy the foreground. Hybrid advertising CGI production offers considerable flexibility, but it also creates more handoffs. Every synthetic element must agree with the plate’s perspective, movement, lighting, color response and optical characteristics.

Classify the visible elements before work starts:

  • Live action: the source plate, actors, environments and captured effects.
  • CG: products, replacement geometry, digital sets and manufactured surfaces.
  • Simulated: particles, fluids, smoke, destruction, cloth and soft-body motion.
  • Practical: in-camera props or effects that may require cleanup, tracking or interaction.
  • Finishing elements: atmosphere, grain, glows and other compositing treatments.

This classification should reflect the final framing. A motion-blurred product occupying a small part of the image does not need the same geometric detail as a hero close-up. Likewise, a distant fluid accent can use a different representation from a foreground simulation whose surface behavior carries the shot. Locking the intended framing early prevents wasted work while protecting the details that communicate the product and concept.

Convert the creative brief into a shot specification

The brief should become a technical specification before asset or FX production. Record the hero action, shot duration, frame rate, framing, aspect ratio, delivery resolution and intended platforms. A master commercial, vertical social crop and product-page loop may require different safe areas and composition priorities even when they share the same scene.

For plate-based work, collect the highest-quality source plate, its frame range, editorial handles and offline reference. Gather camera and lens information where available: sensor or film-back dimensions, focal length, focus changes, camera height and lens-distortion references. A solve may still be possible without complete metadata, but the uncertainty should be documented because it affects perspective, scale and edge alignment.

Reference material should include:

  • Approved dimensions, CAD data, orthographic drawings or other scale references.
  • Material references for roughness, metallic response, transmission, labels and surface variation.
  • Approved design frames, brand colors and product-continuity notes.
  • HDRIs, lighting references, shadow direction and practical-light information.
  • Required collisions, contact points, reflections, refractions and unobstructed brand areas.

Separate creative approval from technical approval. Creative approval covers action, composition, product appearance, mood and timing. Technical approval confirms that the camera, scale, frame range, color pipeline, caches and outputs are suitable for downstream work. A design frame can be approved while its camera solve or simulation setup remains provisional.

The specification should also identify likely effect types—particles, FLIP, Pyro, RBD or Vellum—and the outputs required in compositing. That may include separate product, reflection, shadow, atmosphere and object-matte layers. Defining these needs early gives every department a shared target and avoids discovering during compositing that an essential interaction was never rendered.

2. Prepare the live-action plate

The plate is more than a background. It provides the spatial, temporal, lighting and color reference for the entire shot. Before CG production, determine how the camera moved, how the lens affected the image, where the ground and horizon lie, and which practical elements must be retained, removed or used as holdouts.

Inspect and document the source

Review the plate at its original resolution and frame rate. Look for rolling shutter, exposure or white-balance changes, focus pulls, compression artifacts, stabilization, uneven motion blur and barrel or pincushion distortion. These characteristics influence both the camera solve and the treatment needed to align rendered edges and contact shadows.

Document the source resolution, frame range, frame rate, pixel aspect and any cropping, denoising, sharpening, stabilization or color conversion already applied. Stabilization is particularly important: it changes the relationship between camera movement and the recorded image and should not be treated as invisible preprocessing.

Create clean plates, holdouts and mattes where practical objects must be removed or pass in front of CG. Preparing these elements before final compositing clarifies which parts of the photographed scene must remain visible and reduces emergency roto or paint work later.

Manage lens distortion explicitly

Plate-based work commonly uses an undistorted pipeline. The source is undistorted with a calibrated or measured lens model, and camera tracking and CG rendering take place in that neutral image space. The composite is then redistorted to match the original lens.

The important requirement is consistency. If a camera is solved from an undistorted plate but CG is combined with the distorted source without the matching redistortion, edges can slide even when the solve itself is sound. Keep the lens model, overscan requirements and image-area changes documented and versioned.

Establish color management before rendering

Confirm the plate’s source color space, transfer function and intended working space. Houdini, the renderer and the compositing application should use the same color-management policy and viewing transform. Keep the camera-original media available, and avoid baking temporary display corrections into production plates.

Inconsistent interpretation is a frequent cause of integration problems. A render may be lit correctly in a scene-linear workflow but appear too dark, saturated or contrasty if the plate is interpreted differently in compositing. Color decisions should therefore be made through an identified viewing transform, not an arbitrary viewer setting.

3. Solve the camera and establish scene scale

Camera tracking and matchmove must reproduce perspective and parallax, not merely generate a plausible animated camera. Track stable environmental features rather than reflections, moving objects or heavily blurred details. Then validate the result against features that were not used to create the solve.

Place simple test geometry on visible floors, walls and contact points, then scrub the full shot. Drifting geometry, sliding intersections or inconsistent parallax can indicate weak tracks, an incorrect lens model, rolling-shutter effects or a solve overfitted to one area of the frame.

When publishing the camera, verify:

  • Frame numbering, frame rate and editorial offsets.
  • Focal length, film back, image resolution and pixel aspect.
  • Coordinate-system and unit conventions.
  • Distortion or undistortion state.
  • Ground plane, horizon and useful reference geometry.

A tracked camera does not by itself establish physical scale. Use measured set features or known product dimensions to define scene units. If measurements are unavailable, choose a consistent scale, document the assumption and ensure every department uses it.

This matters especially in a Houdini simulation workflow. Gravity, velocity, particle separation, collision thickness, fluid behavior and cloth constraints all depend on scene units and time. An effect can be visually framed correctly yet behave like a miniature or a giant-scale event if the physical scale is wrong.

Publish a matchmove package containing the approved camera, plate media, distortion data, frame range, scene scale, reference geometry, mattes and color notes. A gray-shaded proxy render can expose camera drift and poor contact points before detailed assets or expensive caches are built.

4. Prepare the product and block the action

Once the camera and plate pipeline are approved, asset and animation work can begin. The goal is not only a correct 3D model; the product must preserve brand continuity, respond properly to light and support the interactions required by the effect.

Validate geometry, scale and materials

Confirm product dimensions, orientation, transforms and origin before building simulations around the asset. Inspect normals, topology and subdivision behavior at the shot camera’s distance. Real product edges catch highlights, so appropriate bevels are often essential; their width should reflect the object’s scale rather than a generic modeling preset.

Separate material regions logically. Painted surfaces, polished metal, glass, labels, rubber and internal components may need independent shaders, render settings and mattes. Preserve approved proportions, label placement, finish and identifying details across all shots and department scenes.

Material choices should be based on final lighting and compositing needs. Roughness, transmission, coat, refraction and absorption affect not only the product’s direct appearance but also how it interacts with the environment. Transparent materials, in particular, need sufficient control over attenuation and reflections to avoid turning gray or disappearing against the plate.

Block timing before detailed simulation

Import the approved camera and confirm the frame range, frame rate, resolution and distortion convention. Block product animation and major effect timing with inexpensive geometry. Evaluate contact points, screen direction, negative space, logo visibility and the relationship between the action and the edit.

Expose art-direction controls rather than burying them inside procedural networks. Useful controls include emission timing, source scale, launch velocity, turbulence, collision response and effect offsets. This makes revisions less destructive and allows the shot to respond to creative feedback without rebuilding the setup from scratch.

5. Build and direct the Houdini simulation

Select the solver according to the behavior the audience must read, not according to which technique is most elaborate. The effect must support the product, composition and edit at the final delivery size.

  • Particles: suited to dust, powder, sparks, dispersed droplets and graphic ingredient effects. Attributes such as velocity, age, scale and color provide direct art-direction controls.
  • FLIP: appropriate when liquid volume, pouring, splashing or surface interaction is central. Particle separation, reseeding, collision quality and boundaries strongly affect continuity and contact.
  • Pyro: used for smoke, vapor and fire. Source fields, temperature, velocity, dissipation and turbulence should support both the intended motion and readable silhouette.
  • RBD: suited to impacts, fractures and rigid destruction. Fracture design, constraints, collision representation and initial velocity determine whether the result feels intentional and mechanically credible.
  • Vellum: useful for cloth, flexible covers and selected soft-body behavior. Rest shape, stiffness, damping, pinning and collision thickness are key controls.

Commercial effects often combine physical plausibility with deliberate staging. A splash may need to reach the product on a specific frame, smoke may need to preserve a clean silhouette, or particles may need to avoid obscuring a logo. Guided velocity, animated sources, masks and controlled time adjustments can direct the result while retaining believable motion.

Develop the effect in stages. Start with low-resolution tests of the source, collisions, timing and overall silhouette. Review through the tracked camera rather than relying on the simulation viewport. Increase resolution only after those fundamentals are approved; added detail will not fix incorrect scale, weak composition or mistimed interaction.

Check the setup before hero caching

Inspect collision volumes and source fields directly. Poor collision geometry can cause leaks, explosive impacts or particles passing through the product. Insufficient substeps may miss fast collisions, but increasing substeps is not a substitute for correcting scale, sources or collision representations.

Check gravity, velocity units, emission timing, turbulence, boundary conditions and solver scale. For RBD and Vellum, inspect constraints and collision thickness; for FLIP and Pyro, inspect source and collision fields instead of judging only the final shaded viewport.

Cache for iteration and traceability

Cache approved stages with names that identify the shot, effect, quality tier and version, such as shot010_FLIP_preview_v003. Keep preview, approved and final caches distinct, and never overwrite an approved result with an experiment.

Store enough metadata to reproduce the result: frame range, scale assumptions, software or tool version where relevant, seed and dependency versions. Lighting can then continue with a stable cache while a separate simulation revision is evaluated. If a high-resolution cache changes timing or silhouette relative to an approved preview, it requires its own review.

For guidance on presenting these stages clearly, see building a Houdini VFX breakdown.

6. Light and render for the composite

Lighting should translate the CG scene into compositing-ready image data while preserving the controlled product presentation expected in advertising. The render must inherit the plate’s lighting logic without sacrificing form, brand color or material readability.

Match light direction and reflection structure

Analyze the plate’s key and fill direction, softness, color, falloff and exposure. Bright windows, practical sources and reflected environments may matter more than average brightness. A fluid requires light that reveals volume and edges; a rigid product often needs carefully designed highlights that describe its shape and bevels.

HDRIs, area lights and reflection cards are tools rather than automatic solutions. An HDRI may create plausible ambient illumination but produce distracting or implausible reflections on glossy packaging. Evaluate reflections against the product geometry and photographed environment. Highlights should explain form, not introduce arbitrary streaks.

If the production is still selecting a renderer, the practical differences covered in Houdini Karma versus Arnold can help frame that decision.

Match motion blur, depth of field and volumes

Use shutter settings consistent with the plate’s apparent blur and the tracked camera. A mismatch can make the product feel detached or leave sharp simulation detail inside an otherwise blurred silhouette. Motion-vector passes can provide flexibility, but they must be generated and interpreted consistently.

Depth of field should follow the camera and focus behavior rather than being added uniformly. Retain enough geometric and shading detail around the intended focus plane, and verify defocus through the same lens and distortion workflow used by the shot.

Volumetric effects require representative motion tests. Adjust volume step size, sampling, density and emission to avoid flicker or loss of fine structure. Judge both static frames and motion-heavy sections; a clean still does not prove temporal stability.

Plan render layers and AOVs around comp decisions

A beauty pass is useful for review but often insufficient for integration. Render only passes with a clear purpose, which may include:

  • Diffuse and specular: for balancing form and highlight intensity.
  • Reflection and refraction: for glossy products, glass and liquids.
  • Volume and emission: for smoke, mist, fire and glowing effects.
  • Shadow and interaction layers: for contact with photographed surfaces.
  • Depth and motion vectors: for controlled atmosphere, defocus and motion treatment.
  • Object and material mattes: for isolating labels, product parts and simulation regions.

Render shadows, reflections and holdouts as genuine scene interactions. A generic multiplied shadow may contain the right category of element but still fail if its direction, softness or contact point disagrees with the plate.

Before publishing, confirm alpha interpretation, premultiplication, color space, channel names, frame padding and motion-vector conventions. Review for flickering highlights, broken refractions, missing shadows, clipped emission and render-farm inconsistencies. Keep simulation, lighting and render versions separate so a shader revision does not imply an unnecessary recache.

7. Integrate the shot in compositing

The final composite tests whether all upstream assumptions agree. The plate, product and simulation must share the same camera, timing, distortion model and color pipeline. If they do not, polishing effects will not make them occupy the same image.

In the Nuke compositing workflow, reassemble the renders with the camera and timing used for lighting. Keep undistortion and redistortion stages visible and documented. This makes it easier to distinguish camera drift from a projection, lens or comp error.

Build depth, contact and interaction

Use holdouts and object mattes to establish front-to-back relationships. Smoke may pass behind one part of a product and in front of another; particles may need to disappear behind live-action objects. Separate product, effects, shadows, reflections and atmosphere where review may require independent control.

Contact involves more than a shadow. Depending on the shot, integration may require reflected light, localized color changes, condensation, displaced debris, splashes or reflections on the product and surrounding set. If an effect has no influence on nearby surfaces, it often appears layered over the scene.

Use depth-based atmosphere selectively. Haze can connect foreground and background elements, but excessive application can flatten the product or contaminate controlled brand colors. Check atmospheric occlusion against both plate objects and CG geometry.

Match the image before adding lens effects

Balance exposure, white point, contrast, black level and saturation while preserving the product’s material response. Then match motion blur and depth of field. Grain, bloom and glare cannot repair an incorrect camera, sharpness or focus relationship.

Inspect edges at full resolution and at delivery size. Transparent liquids, reflective packaging, fine particles and blurred silhouettes can expose premultiplication errors, matte contamination and missing holdouts. Resizing and compression may reveal problems that are difficult to see in an enlarged working view.

Add grain across the integrated image so that CG does not carry a separate noise pattern. Apply bloom, glare, chromatic effects and lens dirt selectively, based on highlights already present in the image. Over-processing can conceal detail temporarily, but it usually makes inconsistent edges and contrast more obvious in final delivery.

For a closer examination of this finishing stage, see the guide to CGI post-processing and compositing.

8. Diagnose common workflow failures

Most failures occur at handoffs rather than within a single node or render setting. The fastest troubleshooting method is to connect the visible symptom to the upstream data that could have caused it.

Visible symptom Likely causes Checks and corrective actions
Product drifts or slides against the plate Weak camera solve, inconsistent distortion, frame offset, film-back mismatch or different camera versions Compare the render and plate with a wipe or difference view. Verify focal length, film back, frame range, transforms and the full undistort/redistort chain.
Object appears to float Incorrect scale or contact point; missing shadow, occlusion, reflection or surface interaction Inspect contact geometry and interaction passes separately. Confirm that shadow softness and direction match the plate.
Simulation is unstable or difficult to direct Incorrect scale, poor collisions, unsuitable sources, insufficient temporal sampling or premature high-resolution caching Inspect source and collision geometry at the intended scale. Test substeps only after validating units, motion and solver inputs.
Effects flicker between frames Changing seeds, incomplete caches, inconsistent frame ranges, unstable sampling or differing render settings Review the full sequence frame by frame. Lock the seed, validate cache coverage and compare settings across render nodes.
CG looks too sharp, clean or saturated Mismatched blur, focus, color transform, grain, edge softness or lens response Check color management before grading. Compare blur and fine detail at delivery resolution, then add plate-matched grain near the end.
Transparent or reflective edges look wrong Incorrect materials, unsuitable environment, missing holdouts, bad premultiplication or insufficient AOVs Inspect reflection, refraction, alpha and matte passes independently. Test edges over light and dark backgrounds.

When a problem first appears in compositing, do not immediately change the model or simulation. Isolate the AOVs, verify plate color and lens treatment, and compare the camera, animation and cache versions with those used for rendering. A late comp adjustment may hide an upstream error only for it to recur in another shot or format.

9. Review, validate and deliver

Review at several scales. A thumbnail reveals composition, silhouette and product readability. Delivery size exposes contact, grain, reflections and edge integration. Frame-by-frame inspection catches simulation flicker, matte chatter, cache gaps, texture popping and isolated render errors.

Track creative and technical approval separately. A shot can be visually approved but have the wrong handles, frame rate or output color space. Conversely, a technically valid delivery can still fail the intended timing or product emphasis.

Final shot checklist

  • Editorial: Confirm resolution, aspect ratio, frame rate, range, handles, retimes and frame offsets.
  • Color: Verify working, display and output transforms. Check brand colors and highlight retention in the required delivery space.
  • Alpha and mattes: Confirm straight or premultiplied alpha, holdouts, object IDs and any required utility passes.
  • Rendering: Validate the approved camera, asset, material, simulation, lighting and render versions.
  • Compositing: Check registration, lens distortion, atmosphere, black levels, motion blur, depth of field, grain and edge contamination.
  • Traceability: Record identifiers such as camera_v###, anim_v###, sim_cache_v###, render_v### and comp_v###.

Validate the exported delivery rather than relying only on the comp script or review movie. Open the final sequence in a clean environment; inspect the first, middle and last frames; verify numbering, handles, color and alpha; and check for missing media or playback errors. The advertising VFX workflow is complete only when the approved image meets the delivery specification and its major dependencies can be traced to their approved versions.