#AI Enhance

One click turns a 3D scene into a photorealistic venue image — without changing your design.

AI Enhance is the feature that closes the gap between a 3D model and a render. You take your scene — exactly as you've designed it, with all your real props in place — and AI Enhance produces a photorealistic image of that same room. Real audience. Real ceiling detail. Volumetric light. Photographic material finish. The geometry, the brand colors, the stage layout, the LED content, the camera angle — all preserved. Only the finish changes.

#What it does

• One-click photorealistic render. Pick an AI mode in the capture panel, click AI Capture, and 20–60 seconds later you have a photorealistic image of the same room.
• Three AI engines, you pick. OpenAI (the Low, Med and Hi modes) is tuned for clean event renders. Google's Gemini image model (the Nano and Nano Pro modes) is the best of the three at keeping your set, your presenters, and your screen content exactly as you built them — and Nano costs less than OpenAI's Med. Flux Kontext (the Flux mode) stays closest to your original and is the most permissive, so it's the one to reach for when another engine refuses something it shouldn't (branded characters, IP-adjacent content).
• One picker, ordered by price. The capture panel has a single menu listing every option cheapest first: Normal (no AI, free), Low (🪙1), Flux (🪙2), Nano (🪙2), Med (🪙3), Nano Pro (🪙8), Hi (🪙10). The credit cost is on the mode button before you spend it.
• Five built-in styles. Photoreal (neutral, true-to-life), Cinematic (dramatic gala lighting), Warm Gala (golden, intimate), Sketch (concept-art style), Neon Cyber (high-tech keynote vibe). Each is a one-click preset.
• Custom prompts. Override the preset with your own prompt for specific creative direction. Pick + New Prompt in the style menu, then Save As to keep it as a personal preset; saved presets get Save, Rename and Del buttons.
• Before / after lightbox. Every enhanced image opens in the lightbox with a slider over the source render — Original on the left, Enhanced on the right. Drag it to see exactly what changed.
• Project Camera Roll. Every capture and enhance lands in a per-project Camera Roll, browsable, shareable, and downloadable. Up to 50 jobs persist across reloads so you can come back later.
• Refusal fallback. If an engine refuses your prompt (content policy), the lightbox marks the shot Refused and offers the other engines as one-click Try with … buttons, most permissive first — usually Flux Kontext, which is the likeliest to accept it. A refused shot costs nothing; the retry reuses your original source, and nothing is charged unless you actually take it.
• Share renderings with your team or the world. Mark any capture as Just me, My team (anyone on your team), or Public on EVT3D (eligible for the EVT3D landing page, with credit to your team, you, both, or Share anonymously — no credit shown).
• Share-link image from captures. Any image in your Camera Roll can be set as the project's share-link image — what Slack or iMessage show when you paste the project's link. Your Home page thumbnail keeps updating on its own each time you save.

#How you use it

1. Frame the shot. Set up your camera the way you'd want the render to look — your chosen angle, your composed scene, your Look applied. Anything visible in the viewport is what AI Enhance will work from.
2. Open the capture panel. Click Capture Rendering in the toolbar.
3. Pick your mode. Click the mode button (the gear) next to Capture — one menu, cheapest first. Low for ideating, Nano when it matters that the set comes back exactly as you built it, Nano Pro or Hi for the final. Flux if another engine refused you. The first time you pick an AI mode in a session, an Enable AI Enhance? dialog lists the costs; click Enable AI Enhance to continue.
4. Pick a style or write a prompt. Pick one of the five presets from the style menu for a one-click look, or write a custom prompt for specific direction ("warm cocktail-hour lighting, golden hour reflections off the floor, full audience standing"). Click Save As to keep the prompt as a personal preset if you'll reuse it.
5. Click AI Capture. With an AI mode picked, the Capture button reads AI Capture. The capture is taken, sent to the engine, and a thumbnail appears in the job strip showing the job's progress.
6. Open the lightbox when it's done. Drag the slider between Original (your 3D capture) and Enhanced (the photorealistic version). Download, share, or use it as the share-link image.
7. Iterate. Don't love the result? Click AI Capture again with a different prompt, style, or mode. Each job is independent — you can run several variations and pick the best.
8. Share or keep. Click the share button on any capture in the lightbox (Share capture). Pick Just me, My team, or Public on EVT3D (with credit choices: your team, you, both, or anonymous).

#What you should know

• AI Enhance preserves the scene. It does not redesign your stage, change your branding, or move your audience seating. The prompts are scaffolded around preservation — the engine is being told what to add (lighting, ceiling, audience), not what to invent.
• Some prompts get refused. AI engines have content policies. If your scene contains a branded character or sensitive imagery and one refuses, the lightbox shows one-click buttons to try the others. OpenAI and the Gemini-based Nano modes filter similarly, so if one of those turns you down the one to try is Flux Kontext — it's the most permissive of the three, and it's offered first for exactly that reason.
• Cost is per-image, in credits. Each AI shot costs 🪙1 to 🪙10 depending on the mode; Normal captures are free. The mode button shows the cost before you commit, and the 🪙 credit meter — in the capture panel and on your Home page — shows what's left in your pool.
• LED content gets re-imagined. AI engines tend to alter what's on a screen. If exact screen content matters (a specific logo, sponsor slide, or graphic that has to be readable), the screen surface may need to be composited back in manually or the prompt steered to "keep screen content exact."

#Why it matters

The traditional path from concept to a render the client can sign off on goes through a render farm: schedule a job, wait hours or days, get back a beautiful image that no longer matches your latest design tweak. AI Enhance shrinks that to less than a minute, on the design you have in front of you right now. The render keeps up with the design, not the other way around — and "show me the gala look" stops being a question that costs you a day of someone else's time.