Blender Character — Near-Infrared (NIR) Skin, Eyes & Modular Accessories for Synthetic Face Dataset -- 2
₹12500-37500 INR
Posted 12 minutes ago
•
Ends in 6 days
₹12500-37500 INR
Paid on delivery
OVERVIEW We are building a synthetic human face dataset in Blender using the open-source Google GNM parametric head extension: [login to view URL] Please note: We will not be sending pre-made .blend files upfront. You can download and install the free Google GNM Blender extension directly and build all materials, eye shaders, hair, and accessories around the standard GNM head setup. Everything you deliver must plug cleanly and easily into our existing Google GNM Blender workflow. Our goal is to use an automated Python rendering pipeline to randomize face shapes, skin profiles, hairstyles, facial hair, glasses, and headwear—generating diverse synthetic human portraits (similar in variety and framing to Microsoft’s FaceSynthetics dataset: [login to view URL]), rendered to match real Near-Infrared (NIR) camera imagery. Target IR Output & Eyeball References: - Sample IR Face Output: [login to view URL] - High-Resolution NIR Eye & Iris Reference: [login to view URL] (These links—along with additional sample IR images we will share with you—serve as our visual target. We will iterate together on your initial sample renders to fine-tune the look so it closely matches our actual camera output.) We are looking for an experienced Blender look-dev / character artist to deliver: 1. An IR-accurate skin, eye, and mouth material library tailored to the Google GNM head. 2. A modular library of hair and accessories that fit cleanly across Google GNM head variations. 3. A calibrated lighting/render setup inside Blender that reproduces real NIR image characteristics. ------------------------------------------------------------ 1. IR-ACCURATE SKIN, EYE & MOUTH MATERIAL LIBRARY - IR Skin Tones & Ages: Distinct skin profiles covering very fair to very deep tones (referencing the Monk Skin Tone Scale) plus age variations (smoother skin for younger faces, fine lines/pores/sunspots for older faces), calibrated specifically for how melanin, subsurface scattering, and skin reflectance behave in monochrome Near-Infrared imagery. - Close-Up Detail & Visible Skin Marks: High-resolution (2K minimum) PBR textures that hold up in tight portrait close-ups—ensuring pores, fine lines, and natural skin marks (moles, freckles, blemishes, scars, and light stubble shadow) remain clearly and accurately visible under IR illumination, with zero seams or stretching on the GNM head mesh. - Script-Friendly Master Shader: A single master skin shader with clearly exposed, easy-to-script controls for skin tone/IR reflectance, roughness/shine, subsurface scattering, freckle/mole/blemish intensity, and stubble shadow, alongside pre-configured presets for all profiles. - IR Eyeball & Iris Physics (Priority): * Eyeballs must accurately replicate real Near-Infrared optical behavior (matching the NIR eye reference link above): - Deep black pupil with a sharp white glint/dot (corneal specular reflection) when looking toward the camera. - Clearly defined grey iris with realistic iris pattern/texture and natural IR reflectance across different iris shades. - Bright white sclera with realistic wetness and a subtle tear line. * Natural IR-tuned materials for lips, eyebrows, eyelashes, and interior mouth parts (teeth, gums, tongue) on the GNM mesh. ------------------------------------------------------------ 2. MODULAR ACCESSORY LIBRARY (COMPATIBLE WITH GOOGLE GNM HEADS) We need a comprehensive library of modular, mix-and-match assets built at real-world scale with clean topology and PBR materials, designed to maximize visual variety and generate a wide range of unique synthetic combinations: - Head Hair & Eyebrows: A wide variety of hairstyles (short, medium, long, straight, wavy, curly, coily, tied back, shaved/buzz, etc.) plus diverse eyebrow shapes and densities in multiple natural shades (including grey and white). - Facial Hair: A broad selection of styles (light and heavy stubble, short beards, full beards, moustaches, goatees, sideburns, etc.) across multiple shades. - Eyewear: A diverse collection of prescription glasses (thin metal, thick acetate, rimless, half-rim) with clear refractive lenses, alongside sunglasses (aviator, square, round, wrap-around, oversized) with tinted, gradient, and mirrored lenses that behave realistically in IR. - Headwear & Face Coverings: A wide assortment of headwear options (caps, beanies, wide-brim hats, headscarves, turbans, hoods, etc.) plus various surgical and cloth face masks. GNM Compatibility & Automation Requirements: - Parametric Fit: Because Google GNM generates varying face shapes, head sizes, and expressions, accessories must adapt/fit cleanly to generated GNM heads with zero clipping or floating gaps (e.g., bound via surface/mesh deformers or rig/shape-key compatible setups so they follow head shape changes automatically). - Clean Layering: Accessories must combine together (for example, wearing glasses, a cap, and a beard simultaneously) without intersecting each other. - Automation-Ready Structure: Because we will randomize combinations via a Python script, all objects, collections, materials, and shader parameters must follow a strict, consistent naming convention and clean hierarchy. ------------------------------------------------------------ TECHNICAL & LICENSING REQUIREMENTS - Built and tested in Blender (Cycles) using native Blender tools and the free Google GNM extension—no paid add-ons required to open or render. - Textures saved with relative paths or packed cleanly into the .blend files. - Usage Rights: All work must be original or properly licensed so we have unrestricted rights to use the 3D assets internally to generate and use synthetic 2D image datasets commercially. ------------------------------------------------------------ DELIVERABLES 1. Master .blend file(s) built on the Google GNM head setup containing the IR-calibrated skin/eye/mouth shaders, lighting/render setup, and sample generated head variations. 2. Organized skin/eye material library and texture folders. 3. Modular accessory library .blend file(s) cleanly structured for Python script swapping and automatic fitting to GNM head shapes. 4. Brief documentation ([login to view URL]) explaining how to apply everything into our Google GNM extension setup, naming conventions, and exposed shader parameters. ------------------------------------------------------------ HOW TO APPLY Please include: 1. Portfolio samples showcasing your Blender look-dev, shading, or 3D asset work (2–3 close-up portrait renders showing realistic skin, eyes, hair/facial hair, or eyewear are given priority if available, though candidates with strong relevant 3D modeling/texturing skills can also apply). 2. A short note on how you will approach the IR skin marks, IR eyeball/iris reflection behavior, and fitting modular accessories to the parametric Google GNM head. 3. Your estimated timeline and cost broken down by milestones—specifically quoting an initial paid milestone batch consisting of IR skin profiles, the IR eyeball setup, 8–10 assets per accessory category, and initial sample renders for iterative alignment with our reference IR images. Once we review and approve this first batch (and confirm any adjustments needed), we will proceed with expanding the full asset library. Please include pricing both for this initial milestone batch and for adding more assets in subsequent milestones.
Project ID: 40823859
About the project
7 proposals
Open for bidding
Remote project
Active 4 mins ago
Place your bid
Bid amount Email addressBenefits of bidding on Freelancer
Set your budget and timeframe
It's free to sign up and bid on jobs
7 freelancers are bidding on average ₹19,619 INR for this job


My Blender experience in 3D modeling, photorealistic rendering, and 3D visualization aligns directly with building the synthetic face assets and NIR-focused render setup you need. I can work around the standard Google GNM head extension and create clean, reusable materials and modular elements that integrate into an automated Blender workflow. Alongside Blender, I work with Maya, 3ds Max, ZBrush, DAZ 3D, KeyShot, CLO 3D, Photoshop, and related modeling tools. This broader character and visualization background supports detailed skin, eyes, hair, facial features, and accessories, while my rendering experience helps refine monochrome appearance, surface response, highlights, and portrait realism against your reference imagery. I can also expose practical shader controls and organize assets for consistent reuse across varied head shapes. I would be glad to contribute a polished, script-friendly Blender library for your NIR face dataset and discuss the project further.


Hello! Your first batch for the GNM head, ready 14 days after award: IR skin profiles, the eye setup, 8 assets for each accessory group and sample renders. You pay us nothing until you've seen it working. Here is what you get: • One master skin shader with scriptable controls (tone / IR reflectance, roughness, scattering, freckles, moles, stubble) and presets from very fair to very deep, young to old • IR eyes: black pupil, sharp white glint, grey patterned iris, bright wet sclera; IR-tuned lips, brows, lashes and mouth interior • 8 hair, 8 facial hair, 8 eyewear and 8 headwear / mask assets that follow different GNM head shapes without gaps or clipping, and stack together (glasses + cap + beard) • Strict naming, a clean hierarchy and a README for your Python randomizer • First preview 24 hours after award: skin, eyes and one render under NIR lighting, to align with your camera images How we work: after the day-1 preview, the first milestone (at least 20%) goes into escrow before the full build and is released only when you approve. Good to know: in NIR, melanin absorbs far less light, so skin tones converge and veins and scattering show more. We calibrate each tone against your reference images instead of simply desaturating a visible-light render. Fitting is tested on several GNM head variations before delivery. Our own demos: a draped wool shawl, a photo-real living room and a game-ready chest. We have no face dataset project to show yet. Price: ₹21,400 for this first batch. Further batches are quoted per group after you approve it. Which NIR band is your camera (850 or 940 nm), and can you send a few sample frames? P.S. Oldrich, the owner, directs the work and stands behind every result. We work with the help of modern technologies, including AI, which keeps us fast and precise.


My 10+ years of experience in 3D modeling and rendering make me well suited to develop the Blender-based visual assets required for your synthetic face dataset. I create accurate, polished 3D models and realistic renderings with close attention to scale, surface appearance, and final image quality. My background in 3D visualization, product design, industrial design, and animation supports the precision needed for modular accessories, clean-fitting head assets, and a consistent calibrated render setup. I can also work with detailed material and lighting requirements to produce believable monochrome NIR-style facial renders, while keeping the assets organized for practical use in an automated Blender workflow. I am quality-focused, reliable, and available for ongoing communication and iteration. Choose me to help build a professional, reusable character and accessory library for your Google GNM dataset.


Hi, **Understanding of the Project** You need a Blender artist to create or modify a character for a Near-Infrared (NIR) visualization project, ensuring the model and materials are suitable for the required rendering, technical visualization, and final output. **My Solution** I can help with the Blender workflow, including: - Character modeling or modification based on your references - Clean mesh topology and proper geometry - Material setup tailored for NIR visualization - Shader adjustments to represent different infrared responses - Lighting, camera positioning, and scene setup - Rendering and preview iterations to validate the visual output - Organized Blender files with editable objects and materials I’ll pay close attention to the technical requirements of NIR rendering rather than relying only on conventional RGB appearance, ensuring the scene is configured according to your desired visualization and reference results. **A Few Questions** 1. Do you already have a character model, or does it need to be created from scratch? 2. Do you have specific NIR material values, wavelength ranges, or reference images to follow? 3. Do you need final rendered images/videos only, or a fully editable Blender project? I’d be happy to discuss the technical details via Freelancer chat and help deliver a well-organized Blender project. Best Regards, Amit


Hello, I’m a 3D artist with 5 years of experience in 3D Rendering, Blender, and 3D Visualization, with a strong focus on character look-dev, shaders, and production-ready asset setups. I’ve understood your Google GNM workflow and the need for IR-accurate skin, eyes, and modular accessories. I would build a clean master skin shader with exposed controls for tone, roughness, SSS, and skin detail, then create an NIR-tuned eyeball/iris setup with realistic glints, sclera response, and tear line behavior. Accessories would be organized with strict naming, clean hierarchy, and fit-tested for the parametric head so they can be scripted and swapped without clipping. I’d keep communication clear throughout the project, share sample renders early, and iterate closely with your reference images. I’d be glad to discuss the best starting batch and align the first deliverables with your dataset needs. Regards


My 3D rendering and 3D visualization skills align directly with your Blender-based synthetic face dataset, particularly the need for realistic, camera-oriented NIR portrait output. I can contribute to developing visually consistent character materials, eye and skin appearance, and render setups that support varied face configurations and close-up imagery. I understand the importance of clean, reusable visual assets for an automated workflow. My focus would be on producing polished 3D results that integrate naturally with the Google GNM head setup, including believable monochrome skin response, facial details, eyes, hair, and modular accessories. This project’s emphasis on realism, variation, and dependable rendering fits my visualization background well. I would be glad to contribute my 3D rendering expertise to this dataset and help create a reliable, visually accurate asset library for your Blender pipeline.


Hi, I read your brief for Blender Character and the focus on creating IR-accurate skin materials caught my attention. I've created similar work, specifically in developing character assets and shaders tailored for unique rendering pipelines. One quick question: how closely do you envision the skin profiles aligning with the Monk Skin Tone Scale in terms of age variations? Happy to chat here if you'd like to talk it through. Kind regards RdF888
内容来自 Freelancer News,版权归原作者所有。
阅读原文 ↗