Rafael Grassetti

Optimizing 3D models for the web and AR used to mean spending hours manually cleaning up topology. I’ve been working on a lightweight tool called Mesh Simplifier to speed up that workflow directly in the browser. In this quick test, I took a 67k-poly bed model and slashed the geometry by 50%. The result? The file size dropped from 9.6 MB down to 1.9 MB..all without losing the essential details or messing up the textures. It makes exporting fast, web-ready .glb and .obj files for AR and WebGL a whole lot less painful. And it’s a free web tool! Try it here: https://lnkd.in/g78AnEQ9 How are you currently handling 3D asset optimization in your pipeline? #3DModeling #WebGL #AugmentedReality #3DWeb #TechArt

9

Facial animation in #iclone8 Reallusion can be done in several ways: • Face Puppet • Face Key • AccuFace • Digital Soul In this short video, I give you a quick overview of how each one works. I’m also sharing some links to tutorials I have on my channel in case you want to dive deeper. https://lnkd.in/dXwxfDvN https://lnkd.in/dyc4Yw54

33

5 Comments

🚀 LTX 2.5 Inpainting HDR Workflow in ComfyUI I've customized and organized the LTX 2.5 Inpainting workflow HDR EXR input and output, aiming to preserve dynamic range throughout the AI inpainting process. I'm sharing a screenshot of the entire workflow to showcase its structure and how I've adapted it for a more VFX-friendly pipeline.🎬 The goal is to bring AI-powered inpainting closer to real compositing workflows, where preserving highlight information and maintaining flexibility for downstream compositing are essential. I'm excited to keep exploring how we can integrate AI tools into traditional VFX without losing the control and image quality we need. More experiments and workflow breakdowns coming soon! LTX.io #ComfyUI #LTX25 #Inpainting #HDR #OpenEXR #VFX #Nuke #AICompositing #GenerativeAI

88

2 Comments

3D Artists you seriously need to get on top of MCP. This week has been incredibly eye opening and fun with how much I've built. Some things I've done with Claude Code, C4D MCP and Claude Chat over this week. Custom soft body rigs and simulations with complete control. Condensation and internal bubble rigs with fine tuning sliders. Illustrator to Photoshop to C4D bridges and plugins. Complete re organisations of massive 100+ product scene files all linking to different materials and lighting setups in one click. Automation over masks, material creation and links. Custom Animations up to 30 seconds long with strict object placements and objectives Model rigs, Xpresso rigs, Cloner operations. Have you been using MCP? If so, how are you using it? If not, why not?

52

29 Comments

ARTISTS, some of the loudest voices giving career advice have never sat in your seat. ▪️ In our industry, you can advise creatives without ever being a creative. ▪️ You can coach job seekers without navigating the current job market yourself. ▪️ You can filter portfolios without ever having to build one under the pressure of student loan debt, deadlines, and repeated rejection. Last week, I reviewed a portfolio from a student who was advised by their university career advisor to "cast a wide net." Broaden their skill sets. Apply to every job they could find. Take whatever position accepted them. The company did not matter. The role did not matter. Just get their "foot in the door" and everything would be fine. That advice may have made sense during a hiring surge. In this 2026 climate, it almost made me faint! Right now, "cast a wide net" can dilute your positioning, confuse your skill narrative, deplete your energy, and send you down a path you will later have to pivot from. 💡Strategy is everything! This market rewards alignment, clarity, connection, and visibility. Not volume. Before you take action on anyone’s career advice, ask yourself: 1️⃣. Have they ever sat where you are sitting? Have they applied for jobs in this market? Built a portfolio under pressure? Navigated rejection cycles recently? 2️⃣. Can they show outcomes? Not inspiration. Not theory. Actual people they guided from early career to employed, without making a phone call on their behalf.. That distinction matters. If their advice alone could not stand on its own without them leveraging a personal relationship with the hiring manager, then their strategy did not get the person hired. Their access did. 3️⃣. Are they advising from today’s realities, or from their memory? Listen closely to their own journey and the picture they are painting of what they think yours will look like. 4️⃣. As controversial as this topic has become, are they considering the unique challenges you may face because of the body and identity you exist in? This deserves its own post, because it is not discussed nearly enough in hiring and advising spaces. In 2009, my professor once told me mobile compatibility was irrelevant for creative portfolios. He was thinking about how he reviewed work. He was not thinking about where the industry was going. The dangerous part was that, to my inexperienced self, that advice sounded reasonable. Recruiters, coaches, and advisors can outline steps and tell you what hiring managers asked them to look for. That is valuable. But that does not automatically mean they can give you the best strategy to get hired. Please use discernment! Early career talent, you are working harder than many of us have seen in years. You deserve advice that reflects that reality. #MyAnimationLife #AnimationJobs #AnimationAdvice #AnimationIndustry #AnimationCareer

25

3 Comments

Anyone else feel like lookdev takes longer deciding than actually doing? I keep running into this loop: tweak → render → wait → forget what I was testing → tweak again. I’ve been messing with the Live Render node in Houdini 21 to see if it helps reduce how often I break flow during lookdev, especially when dialing materials and lighting together. It’s not a magic switch, but it does change the iteration rhythm inside Solaris. I ended up recording a breakdown of how I’m using it in a simple setup. Link’s in the comments if you’re curious. Curious how others are handling lookdev iteration right now.

5

1 Comment

Recently I built a small performance HUD overlay for Houdini that shows CPU, RAM and VRAM usage directly in the viewport. The idea is simple: instead of opening profiling tools, you always have a bit of performance telemetry visible while you work. The HUD runs outside the Houdini GUI thread, so it keeps updating in real time even while Houdini is busy cooking and the interface is blocked. That turns out to be surprisingly useful during longer cooks, because you can still see what the current Houdini instance is actually doing. While working recently I noticed something unusual during a Delta Mush cook just by glancing at the overlay, as shown in the video. That led me to investigate further, and I ended up finding a simple workaround that made the operation dramatically faster and reduced memory usage. Sometimes just having the right information visible at the right moment is enough to reveal things you would otherwise miss. This HUD will be part of Houdini Supercharged extension.

374

7 Comments