Skip to content
--:-- MST

Portfolio · 2026

I turn visions into experiences people remember.

01 · About

A craftsman
of digital
worlds.

Name
Reese Nelson
Location
Provo, Utah
Available
Currently Available
Education
UVU '26
Reply time
Within 24h
Remote
Worldwide

Hey! I'm Reese Nelson — a UVU grad with a passion for building immersive digital worlds. I specialize in scene environments, VR simulations, and interactive experiences that blend technical precision with creative storytelling.

The work here is a mix of freelance projects and industry collabs — each built around thoughtful design, clean workflows, and experiences people actually remember. My main toolkit is Unreal Engine, Blender, Substance Painter, and DaVinci Resolve.

Whether I'm crafting a detailed environment, designing a VR experience, or untangling a gnarly technical problem — I love bringing ideas to life in ways that feel immersive, intuitive, and visually compelling.

🎮

Origin story

Started taking classes in high school about computers

🌊

Weirdest project

Simulated a storm with pure math

⏱️

Hours logged

UVU Engineering Building has over 1991 hours of team effort.

✈️

Remote ready

Works with teams worldwide

🏛️

Proudest work

Digitally preserved a Greek church

Freelance Environment Artist

2024 -- Now

Independent · Provo, Utah · Remote

Unreal Engine 5VR / XRArchvizReal-time

UVU Walk-through Project

2025

UVU Innovation Lab · Lead Artist

Unreal Engine 5Meta QuestLumen GI

3D Generalist · Personal

2022 -- 2024

UVU Digital Media · Provo, Utah

BlenderSubstance PainterCycles
Last updated May 2026 Available

02 · The Pinboard

Drag, arrange,
make it yours.

A live pinboard of projects, memos, sketches & more. Grab anything -- move it anywhere.

100%
VR
UVU Engineering
3D
Luxury Apartment
Sim
Ocean Storm
Scan
G.A.P. · Greece
VR
J.E.F.F.

Note to self

Every model has a place

Hire me for

VR sims · Archviz · 3D scanning

Currently

Building another enviroment using the new Blender 5.0 verison

Project Brief · 001

UVU Engineering

Built in Unreal Engine 5. Stereoscopic VR tour with Lumen GI. Meta Quest. 11,713 objects.

Project Brief · 002

J.E.F.F.

3 military aircraft · 5 missile systems · Educational VR experience

Fun fact

J.E.F.F is controlled through Hand-Tracking

Currently

I am learning more about Claude and how to implent it.

Collab

I have collabed with UNESCO and UVU Engineering Department

Hobbies

I love to bake, qulit and draw outside of 3D developing

Project Brief · 003

G.A.P. · UNESCO

LiDAR + Gaussian Splatting · Agios Achilios, Greece · Cultural preservation for UNESCO

Sim
Storm at sea

Business Card

Reese Nelson

3D Artist · XR Developer

reesenelson125@gmail.com

reese-nelson.com

Provo · Utah · USA

Drag any note -- pin it wherever you like.

03 · The Archive

Selected
work, 2024-26.

Click any project to open the full case study.

+

More projects
coming soon

04 · Now playing

Demo Reel '26.

A tightly-cut 90-second showreel spanning 2024-26 -- real-time captures and offline renders from every project above, color-graded in DaVinci Resolve.

Demo Reel '26 — a 90-second, music-only showreel of real-time captures and offline renders. No dialogue or narration.

05 · Methodology

From concept
to final pixel.

Every project runs the same four phases. Here's the Ocean Storm simulation — from a blank viewport to a real-time storm at 60fps.

Phase 01

Reference
& Discovery

Before opening a single tool, I build a visual language — mood boards, reference pulls, material studies. The goal is a shared direction so there are no surprises at the end.

PureRefMood BoardsStyle Sheets
01
Research
Room Final Mesh
02

Phase 02

Block-out
& Mesh

Rough geometry first — no textures, no lighting. I lock in scale, flow, and camera composition with clean topology before committing to detail. Measure twice, model once.

BlenderPoly ModellingUV Unwrap

Phase 03

Lighting
& Materials

PBR materials and Lumen GI are dialled in simultaneously. Every material decision is tested under the final lighting so nothing looks good in isolation but wrong in context. The evening pass is the hardest to get right. The image shown is from the project.

Substance 3DLumen GIPBR Maps
03
Material & Lighting
Final Beauty Render

Phase 04

Beauty Render
& Delivery

Final path-traced beauty render, colour-graded in post. All render passes exported and ready to finilize.

Path TracingColour GradeRender Passes

06 · Final chapter

Let's make
something real.

What are you building?

When do you need it done?

Case Study · 01 · VR Simulation · Project B.R.E.A.D.

Smith Engineering & Technology Building.

Client

UVU · Dean Flanagan

Team

5 Members

Timeline

Dec 2023 - Mar 2025

Platform

Meta Quest

01 · Overview

A building before it exists.

This virtual reality tour provides an immersive experience of the upcoming Utah Valley University Smith Engineering and Technology Building. Students and faculty gain insights into the building's size and the comprehensive amenities it will offer -- from the Forum and Drone Lab to the Machine Shop and Lecture Hall.

Dean Kelly Flanagan contributed financial support to the engineering building virtual experience, along with the necessary equipment. in collaboration with the Digital Media Department. This space serves as the students' workspace, enabling the realization of the project.

39Production Sprints

Dec 2023 -- Mar, 2026. 28 days, 39 sprints.

tap to flip back

49Pages of Research

Documentation that earned the $50K funding.

tap to flip back

46Donor Rooms

Each room named after a donor to the real building.

tap to flip back

$4MDonated to Building

The virtual walkthrough directly inspired $4 million in donor contributions to the real building.

tap to flip back

02 · Gallery

Wide Shot

Forum (Nu Skin)

Lab

ECE Drone Lab

Workshop

Machine Shop

Interior

Lecture Hall (Spid3rThink)

Lab

Smart Grid Lab

03 · The Team

Meet the team.

Reese Nelson · Kael Harrison · Melia Masek · Jordan Hunter · Brandon Jorgensen

04 · Trailer

Watch the walk-through.

Trailer for the UVU Smith Engineering Building walk-through — real-time render footage, no dialogue or narration.

05 · Details

By the numbers.

The original product file contained 21,152 3D objects, optimized down to 11,713 in the final build. The team completed 39 production sprints across 61 builds, with 12 faulty builds caught during QA and 19 students who play-tested the experience.

This was the first-ever virtual tour of this kind at UVU -- a pioneering project that set the standard for immersive campus experiences. Donors contributed $4 million to the building, with 46 donor rooms featured in the virtual experience.

Want something similar?

Let's build your world.

VR walkthroughs · Architectural visualization · Real-time environments

Case Study · 02 · Real-time Archviz

Luxury ApartmentSuite.

Type

Personal

Engine

Unreal Engine 5

Year

2025

Lighting

Lumen GI

01 · Overview

A penthouse brought to life.

A luxury penthouse interior built in Unreal Engine 5 with Lumen global illumination. Features marble columns, custom hi-fi speaker systems, floor-to-ceiling windows, designer furniture, and accent LED ceiling lighting -- all rendered in real-time at interactive framerates.

Unreal Engine 5Lumen GIPBR MaterialsArchvizCustom Props
Got a project in mind?Let's talk →

02 · Gallery

Scroll to explore · Click to expand

03 · Flythrough

Walk through the space.

Flythrough of the apartment interior — real-time render footage, no dialogue or narration.

Quick facts

🏠Every prop was hand-modelled — from speaker grilles to couch cushions.
💡Ray-Traced Path calculates indirect light bounces in real-time — no baking.
✦The Luxury Apartment Suite is all made in Blender with dynamic lighting.

Let's build your space.

Got a project in mind?Let's talk →

Case Study · 03 · Real-time Simulation

Ocean StormSimulation.

Type

Personal

3D Modeling

Blender

Year

2026

Type

Simulation

01 · Overview

Procedural chaos.

A real-time ocean storm environment featuring lighthouses, buoyancy physics, volumetric rain, and dynamic wave displacement. Built in Blender with foam generation, underwater caustics, and storm lighting.

Got a project in mind?Let's talk →

02 · Gallery

Night

Storm · lighthouse beams

Wireframe

Wave mesh topology

03 · In Action

Watch the storm.

Ocean storm simulation in motion — no dialogue or narration.

Wireframe Pass

Wireframe pass of the same ocean simulation — no dialogue or narration.

Quick facts

🌊Wave displacement uses a custom shader — no physics engine, pure math.
💡The lighthouse beam is a real-time volumetric light that cuts through rain particles.
⚡Runs at 60fps — a mixed of baked and real-time rendering.

Let's simulate.

Got a project in mind?Let's talk →

Case Study · 04 · Cultural Heritage · G.A.P.

Agios AchiliosPreservation.

Client

UNESCO

Tech

LiDAR + Gaussian

Location

Greece

Engine

UE5

01 · Overview

Preserving history.

The Gaussian Acquisition & Preservation project digitally captures and preserves a Greek Orthodox church on the island of Agios Achilios in Greece. Using LiDAR scanning and Gaussian splatting technology, the church's centuries-old frescoes, ornate woodwork, and architectural detail were captured as high-fidelity 3D point clouds and rendered in Unreal Engine 5.

LiDAR ScanningGaussian Splatting3D ReconstructionUnreal Engine 5Preservation
Got a project in mind?Let's talk →

02 · Process & Results

Comparison

Gaussian Splat vs Point Cloud

Scanning

Mobile scan preview

Reference

Church interior

On-site

Mobile LiDAR capture

Equipment

Tripod scanner setup

03 · Video

See the capture.

3D scan capture of the G.A.P. Agios Achilios site — no dialogue or narration.

Quick facts

📡LiDAR + Gaussian splatting — a first in cultural preservation for this site.
🏛️The church dates back to 1774 and holds priceless Byzantine frescoes.
🌍Scanned on-location in Agios Achilios, Greece using the X-Grid Camera.
✦The final render is indistinguishable from a photograph at human viewing distance.

04 · UNESCO Report

Read the newsletter.

The February UNESCO newsletter documenting the Agios Achilios G.A.P. project — from on-site scanning to the final Gaussian splat deliverable.

📄

UNESCO · February 2026

G.A.P. Project Newsletter

Official newsletter covering the Agios Achilios digital preservation project — scanning methodology, Gaussian splatting results, and cultural heritage impact delivered to UNESCO.

Click to view inline ↗

Let's preserve.

Got a project in mind?Let's talk →

Case Study · 05 · VR Simulation · J.E.F.F.

J.E.F.F.VR Simulation.

Team

Two Memebers

Timeline

Apr 2025 -- Apr 2026

Hours

Around 5000

Type

Educational VR

01 · Overview

Learn about military aircraft.

An educational VR experience that lets users explore 3 military aircraft and 5 advanced missile systems in a realistic hangar environment. Built by Reese Nelson and Vryan Arias over approximately 5000 development hours. The experience is interactive, immersive, and designed to teach users about military aviation technology through hands-on VR exploration.

Got a project in mind?Let's talk →
UE5Engine

Built entirely in Unreal Engine 5, hand-tracking driven.

tap to flip back

3Military Aircraft

Modelled from reference, not scan -- original 3D.

tap to flip back

5Missile Systems

Each system is fully interactive in VR.

tap to flip back

Q3Meta Quest 3

Hand-tracking on Meta Quest 3 — no controllers, fully gesture driven.

tap to flip back

02 · In Action

Watch J.E.F.F. in action.

J.E.F.F. VR training sim in action — no dialogue or narration.

J.E.F.F. · In Action

A further in-action clip of the J.E.F.F. VR training sim — no dialogue or narration.

Let's build your sim.

Got a project in mind?Let's talk →

Case Study · 06 · UX Research

UVU Admissions Untangled.

Client

UVU · Admissions

Team

6 Researchers

Timeline

Fall 2024

Role

UX Researcher

01 · Overview

A site with too many doors.

UVU's Admissions pages had grown into a maze -- content spread thin across dozens of interlinked pages that sent new, current, and transfer students in circles instead of toward what they actually needed. As part of a six-person research team, I helped run a semester-long usability study to find out exactly where people got lost, and why.

The brief was research only -- no redesigns, no wireframes. Just enough evidence, gathered through surveys, think-aloud sessions, eye-tracking, and card sorting, to hand UVU's development team a clear, defensible path forward within the university's existing design system.

6Researchers

A six-person team split across three research tracks.

tap to flip back

70Page Report

The final findings report, handed to UVU's development team.

tap to flip back

4Research Methods

Survey, think-aloud testing, eye-tracking, and card sorting.

tap to flip back

200+Cards Sorted

Over 200 site sections, split into three 50-card sorts.

tap to flip back

02 · What we ran

Survey Think-Aloud Testing Eye-Tracking Card Sorting

03 · Findings

What we learned.

Across every method, the same story repeated: the information was technically on the site, but never where people expected it. Think-aloud participants struggled to navigate the homepage and would drop into Google before trying UVU's own search. Text that was too small or unstyled got skipped entirely -- and once someone lost the first two sections of a page, everything below it went unread.

Eye-tracking told the same story from a different angle. Students expected tuition deadlines to live on the tuition page and financial-aid contacts to live under "financial aid," not scattered across the Student Life Center and Alumni pages -- a gap that cost one prospective high-schooler the most time of any participant, since she wasn't yet familiar with UVU terms like alumni.

04 · Survey Snapshot

Before we watched anyone click.

ORGANIZED FIND IT LOOKS READABLE

"What matters most on a website?" -- finding content easily beat visual appeal by a wide margin.

NEVER 1-2X/SEM WEEKLY FEW DAYS DAILY

"How often do you use the UVU website?" -- the single largest group of respondents said never.

05 · Recommendations

Handed to the dev team.

Rename the links

Card-sort participants had no idea what tabs like "Adult Learners" meant, or how they related to Admissions.

Merge near-duplicates

Fewer, better-labeled groupings beat an ever-expanding list of near-identical categories.

Top tasks, one click

The most-requested information should never take more than a single click from Admissions.

Surface registration

Changing a legal name currently requires finding the search bar first -- registration needs its own visible link.

06 · The Team

Six researchers, one report.

Reese Nelson · Micah Brailsford · Kobey Chacon · Kenzie Delano · Nathan Ferrell · Jade Xa

07 · Conclusion

Handed off, not shipped.

The scope was research, not redesign -- so the deliverable was a 70-page findings report, not a new website. It gives UVU's development team an evidence-backed map of exactly where students get lost and why: simplify the navigation, cut the number of tabs, rename what's unclear, and put the registration page where people actually look for it.

08 · From the research

Pages pulled straight from the 70-page findings report -- the survey chart, a think-aloud write-up, and an eye-tracking heatmap. Click to expand.

09 · Research Documents

Read the full report.

📄

01 · Fall 2024

Project Brief

Team, problem statement, objectives, scope, and budget for the semester-long study.

Click to view inline ↗

📄

02 · Baseline attitudes

Survey Results

What students said mattered most on a website, gathered before any usability testing began.

Click to view inline ↗

📄

03 · Moderated sessions

Think-Aloud Testing

Where participants got lost navigating the site, in their own words.

Click to view inline ↗

📄

04 · 5 participants, 5 tasks

Eye-Tracking

Gaze-mapped task attempts, including the tuition-due-date task nobody could finish.

Click to view inline ↗

📄

05 · 200+ cards, 3 groups

Card Sorting

How students naturally grouped over 200 site sections, split across three research tracks.

Click to view inline ↗

📄

06 · Handed to the dev team

Recommendations

Rename the links, merge near-duplicates, surface top tasks, and expose registration.

Click to view inline ↗

📄

07 · Wrap-up

Conclusion

The 70-page findings report in summary, and the evidence-backed map handed off to UVU.

Click to view inline ↗

Want research like this?

Let's find what's actually broken.

UX research · Usability testing · Information architecture

Case Study · 07 · UI/UX Design

reese-nelson.com Designed in the open.

Client

Self-directed

Team

Solo

Timeline

2025-26

Role

Designer & Developer

01 · Overview

A portfolio that works like the work.

Most 3D and XR portfolios show polished renders inside a template that feels nothing like the work itself. I wanted the opposite -- a site with the same layered, tactile, slightly theatrical feel as a walkthrough or an environment build, built entirely by hand rather than dropped into a page builder.

Everything you're looking at right now -- the corkboard of dragging sticky notes, the tilt-on-hover project cards, the swipe-between-case-studies navigation, the inline PDF viewer for research documents -- lives in a single self-contained HTML file, styled and scripted from scratch.

02 · Process

From messy desk
to working system.

The starting point wasn't a wireframe -- it was the corkboard. I kept coming back to the image of a designer's actual pinboard: reference pulls, a couple of polaroids, sticky notes at slightly wrong angles. That messiness felt more honest than a clean grid of thumbnails, so I built the About section around it first and let the rest of the site's tone follow from there.

From there it was one project page at a time. The Smith Engineering case study came first, built as a one-off. As soon as a second case study needed the same hero-plus-gallery-plus-CTA shape, I pulled that structure out into a reusable pattern instead of copy-pasting and drifting -- so every project after that (the apartment, the ocean sim, the UX research write-up, this page) slots into a system that already knows how to hold it.

Mobile wasn't an afterthought pass at the end. The pill-style nav collapses into a numbered slide-out menu, project pages support left/right swipe the way a native app would, and the corkboard's drag interactions were rebuilt to work with touch from the start rather than patched in later.

03 · A look at the real thing

About

The pinboard -- reference pulls, polaroids, sticky notes at slightly wrong angles

Work

Filter pills, tilt-on-hover tiles

Two screens off the live site: the corkboard that set the tone in the Process section above, and the Work grid every case study -- including this one -- lives inside of. Click to expand.

1HTML File

No framework, no build step -- every page, style, and interaction lives in one file.

tap to flip back

9Case Studies

From VR walkthroughs to UX research to this page you're reading right now.

tap to flip back

5Core Colors

A tight palette -- void black, two purples, and one hot accent -- used everywhere.

tap to flip back

0Templates Used

Every section, from the corkboard to the lightbox, was built from scratch.

tap to flip back

04 · How it's built

Vanilla HTML / CSS / JS CSS Custom Properties Inline SVG Data Viz Touch / Swipe Navigation

05 · Design principles

Structure with personality.

Type does two jobs on purpose: an italic serif display face carries the emotional beats -- names, hero titles, pull quotes -- while an uppercase monospace handles labels, eyebrows, and anything structural. Body copy stays in a plain sans so long-form case-study writing never fights for attention.

Every project opens into its own full case-study page -- a hero, a two-column overview, flip-to-read stat blocks, and a documents or gallery section -- reused across all nine pages so new work slots into a pattern that already exists, rather than a one-off layout every time.

06 · The Palette

Five colors, everywhere.

Void

#100820

Grape

#5b3aa3

Lavender

#b8a4e3

Hot Accent

#d97ff5

Bone

#e8e0f2

One CSS variable per color, referenced everywhere from buttons to chart bars -- change the five values here and the whole site restyles.

07 · Decisions & trade-offs

Why it's built this way.

Why one file?

A page builder or CMS theme would've fought me on every custom interaction. One HTML file means the corkboard, the tilt cards, and the swipe nav can all touch the same state without wrestling a framework -- the trade-off is I keep the file organized by hand instead of letting a bundler do it.

Why hand-roll every interaction?

Off-the-shelf sliders and lightboxes all move the same way, so nothing feels specific to this work. Writing the tilt, the drag-physics on the notes, and the case-study lightbox myself is slower, but it means the site moves the way I want it to, not the way a plugin decided it should.

Why swipe between case studies?

On desktop the nav and "next project" button are enough. On a phone, swiping between projects reads more like flipping through a lookbook than clicking through a website -- so the same case-study pattern got a second, touch-first way to move through it.

Why a corkboard, specifically?

Most "About" sections are a headshot and three sentences. I wanted the entry point to feel like walking into a workspace instead -- so it's pinned notes and polaroids you can drag around, not a static bio block.

08 · Conclusion

Never really finished.

This site is a living project -- the case study you're reading was added the same way every other project gets added: a new page in the pattern, a new tile in the grid. It grows every time there's new work worth showing.

Want a site like this?

Let's design something that feels like you.

UI/UX design · Front-end build · Design systems

Case Study · 09 · UI/UX Design · VR Onboarding

Game UI Design.

Client

UVU · Smith Eng. VR

Team

VR Project Teams

Timeline

2025-26

Role

UI/UX Designer

01 · Overview

Teaching people to move in VR.

Building the environment is only half the job on a VR project -- the other half is the handful of screens standing between someone putting on a headset for the first time and actually being able to walk around. This case study covers the UI/UX work I've done for movement, onboarding, and in-headset settings across the VR projects I've worked on, starting with the system built for the Smith Engineering & Technology Building walkthrough.

A lot of the people testing that walkthrough -- donors, faculty, prospective students -- had never used a headset before. So the design problem wasn't "make a menu," it was "get a first-time user from putting the headset on to comfortably exploring the building in under a minute, without a live demo standing over their shoulder."

02 · The onboarding flow

Scroll to explore · Click to expand

The full onboarding board: an intro screen with a circular map dial, a choose-your-navigation-mode screen, teleportation and locomotion tutorial cards, an in-headset settings panel laid over a top-down map, a conclusion screen, and the Movement Type selection screen taken through a few iterations. Click any card to expand.

03 · How it works

Teleportation Mode Locomotion Mode Controller-Accurate Tutorials In-Headset Settings Panel

04 · Two ways to move

Comfort is a choice, not a default.

Right after the intro screen, users choose between teleportation (point, click, blink to the next spot -- near-zero motion sickness) and locomotion (smooth movement with snap-turn, sprint, and jump -- more natural, but not comfortable for everyone). Each mode gets its own short, icon-led tutorial card sequence showing the exact controller buttons involved, rather than a wall of instructional text no one reads in a headset.

The choice doesn't lock anyone in. A settings panel, laid over a top-down map of the building, lets a user switch movement modes mid-tour if teleporting feels too disorienting or locomotion is making them queasy -- so the first decision someone makes thirty seconds into the experience isn't a permanent one.

05 · Screen by screen

Every card teaches one thing.

The welcome screen pairs a photo of the actual building with a small circular map dial -- a first, low-stakes look at the space and how it's oriented, before anyone's asked to move through it. From there it's straight to the fork: teleportation or locomotion.

Locomotion gets six tutorial cards -- Movement, Snapping (snap-turn instead of smooth rotation, for comfort), Menu, Selecting, Sprint, and Jumping. Teleportation gets five, dropping Sprint since it's a point-and-blink mode with no run button to teach: Movement, Snapping, Menu, Selecting, and Jumping. Every card follows the same shape -- a controller-hand icon showing the exact button, plus one short instruction bubble, never more than that.

The Movement Type screen -- the radio-button choice between the two modes -- went through three passes before it shipped, tightening the copy and the visual weighting each time so the trade-off (comfort vs. naturalism) reads at a glance instead of needing a tooltip.

06 · Design principles

Designed for first-timers.

Every screen in this flow was built around one constraint: most people using it had never worn a headset before, and no one was going to explain the controls to them in person. That meant leaning on icons and controller diagrams over paragraphs of text, keeping each tutorial card to a single idea, and defaulting new users toward teleportation -- the lower-risk option -- while still making locomotion easy to reach for anyone who wanted it.

It also meant designing for the moment things go wrong: if someone picks the wrong comfort mode, the fix has to be one glance at a map and one button, not a restart of the whole tour.

Have a UI project?

Let's design the interface.

UI/UX design · Interaction design · Design systems

Reference · 08 · Format Research

Choosing the right format, every time.

Type

Independent Research

Docs

4 References

Sources

~42 Papers

Year

2026

01 · Why this exists

The part of the process nobody sees.

Knowing which codec, container, or export setting actually fits the deliverable rarely makes it into a portfolio -- but it's the difference between a render that looks right in review and one that breaks on the client's platform. These are four condensed references, synthesized from roughly 42 papers, specs, and comparison tests, covering audio, image, video, and 3D/XR file formats.

4Formats Compared

Audio, image, video, and 3D/XR -- every surface a deliverable eventually has to cross.

tap to flip back

~42Papers & Specs Reviewed

Condensed into four references, so the takeaways fit on a single read each.

tap to flip back

5Streaming Platforms Tested

Part of the audio blind listening test -- codecs compared across devices and services.

tap to flip back

3Delivery Services Compared

Netflix, Disney+, and Hulu -- checked against the video format recommendations.

tap to flip back

02 · Quick reference

The short version, if you're in a hurry.

The full references go deep -- this is the cheat sheet. General-purpose starting points for each category; the linked PDFs below cover the exceptions and the testing behind them.

🎧

Audio

StreamAAC -- efficient, near-universal support.
ArchiveFLAC -- lossless, still a fraction of WAV's size.
AvoidMP3 past a quick draft -- lossy by design, no longer the best-supported option it once was.
🖼️

Image

WebWebP -- smaller than JPEG/PNG at comparable quality.
ArchiveTIFF or high-quality PNG for anything that may be reprinted.
AvoidJPEG for sharp edges or text -- compression artifacts show immediately.
🎬

Video

DeliverH.264/MP4 -- plays everywhere, no exceptions.
BetterHEVC or AV1 where the platform supports it -- real bitrate savings.
AvoidProRes or uncompressed as a final deliverable -- keep those for editing masters.
🧊

3D / XR

DeliverglTF -- built for real-time and XR, compact and widely adopted.
InterchangeFBX or USD when moving between DCC tools.
AvoidOBJ for anything animated -- no skeletal or animation support at all.

03 · How the testing worked

Real tests, not just spec sheets.

The audio comparison was a blind listening test across three playback setups -- studio monitors, a pair of AirPods, and a phone speaker -- run against the same master encoded at matched bitrates on five streaming platforms. The gap between AAC and MP3 was audible on monitors and nearly gone on a phone speaker, which is exactly the kind of thing a spec sheet won't tell you.

Video went through the opposite test: the same source master, delivered through Netflix-, Disney+-, and Hulu-style transcode pipelines, then compared frame-by-frame at matched bitrates for banding, macroblocking, and motion artifacts. Image and 3D/XR formats were tested more directly -- repeated compression passes for fidelity loss, and round-trip exports between DCC tools (Blender, Substance Painter, Unreal Engine 5) to check what geometry, material, and animation data actually survives a format change.

04 · At a glance

If it's this, use that.

A lookup table for the deliverables that come up most in 3D, XR, and UI/UX work -- built from the same testing above, not guesswork.

🎥

Client walkthrough video

H.264 / MP4

🕶️

Real-time VR / XR asset

glTF

🗂️

Archival 3D scan

USD or FBX

🖨️

Print-ready render

TIFF

🌐

Website hero image

WebP

🎙️

Portfolio narration audio

AAC (stream) / FLAC (archive)

📱

UI icon set / interface art

SVG, or WebP for raster

🧩

Handoff to another DCC tool

FBX or USD

05 · Myths worth retiring

Things everyone assumes.

A few beliefs that come up constantly and don't hold up once you actually test them.

✕

Myth

A bigger file always means better quality.

✓

Fact

Bitrate and encoder quality matter more than size -- a poorly encoded large file can look worse than a well-encoded smaller one.

✕

Myth

PNG is always the safest, lossless choice.

✓

Fact

PNG is lossless, but it balloons in size for photographic detail -- WebP or a careful JPEG often preserves what the eye sees at a fraction of the weight.

✕

Myth

Upscaling a compressed file fixes the compression.

✓

Fact

Resolution and compression are separate problems -- upscaling a low-bitrate source just makes the existing artifacts bigger, not sharper.

✕

Myth

If the file opens, the format choice didn't matter.

✓

Fact

Opening isn't the same as surviving intact -- OBJ silently drops animation data, MP3 silently discards frequencies, and neither one warns you when it happens.

06 · Glossary

The vocabulary, briefly.

The handful of terms that make the rest of this page make sense.

Codec

The algorithm that compresses and decompresses the actual data -- H.264, AAC, FLAC. Different from the container that holds it.

Container

The file wrapper holding codec data plus metadata -- MP4, MOV, WAV. One container can hold several possible codecs inside.

Lossy

Compression that discards some data permanently to save space -- JPEG, MP3, H.264. Usually fine in moderation, compounds badly with re-encoding.

Lossless

Compression that preserves every bit exactly, reversible with no quality loss -- PNG, FLAC, TIFF. Costs file size in exchange.

Bitrate

How much data is spent per second of audio or video -- the single biggest lever on the trade-off between quality and file size.

Interchange format

A format built specifically for moving assets between different tools without losing data along the way -- FBX, USD, glTF.

07 · The full references

Four formats, condensed.

🎧

01 · Codecs & streaming

Audio Formats

MP3, AAC, FLAC, ALAC & Dolby Atmos compared, plus a blind listening test across three devices and five streaming platforms.

Click to view inline ↗

🖼️

02 · Raster, vector & export

Image Formats

JPEG, PNG, GIF, WebP & JPEG 2000 evaluated for compression and fidelity, with Photoshop export methods compared.

Click to view inline ↗

🎬

03 · Codecs & platforms

Video & Streaming

MOV, MP4, H.264, HEVC, VP9 & AV1 explained, with a Netflix, Disney+, and Hulu delivery comparison.

Click to view inline ↗

🧊

04 · Interchange & delivery

3D / XR Formats

FBX, OBJ, glTF, USD & STL compared for geometry, materials, animation, and real-time/XR readiness.

Click to view inline ↗

Want something similar?

Let's build your world.

Research · Documentation · Format strategy