Recommended
Recommended
More Related Content
Similar to Gdc15 michael pavlovichupload
Similar to Gdc15 michael pavlovichupload (8)
Recently uploaded
Recently uploaded (20)
Gdc15 michael pavlovichupload
- 1. Blurring the Separation Between Concept and Production Michael Pavlovich Sr Character Artist, Certain Affinity
- 4. Proving the Manual Process
- 5. ●Quick High res source with Material IDs
- 7. Rough Sketch
- 8. Refined Sketch ● Answers enough questions?
- 9. Component Modeling ● Faster Material ID assignment and iteration ● More Precise Refining
- 10. Sketch Refine ● Answers all (or enough) form and functionality questions for most objects
- 11. Forms Refined
- 12. Secondary Detail
- 13. More Detail
- 14. …More Detail ● Populate your asset libary
- 17. Final High Res ● High res bake model ● Material IDs applied (polypaint or material)
- 18. ●Quick Game Res with Uvs
- 19. ●High Res
- 23. ●High Res ●Dynamesh/Remesh ●Decimate ●Attract Seams by AO ●UV
- 25. ●Source Maps ● Material ID ● AO ● Curvature ● Norm Tangent ● Norm World ● Position ● Thickness
- 26. ●Quick Material Assignment ●Automatic Textures
- 28. ●Output Maps ● Albedo ● Roughness ● Metallic ● Cavity ● Specular ● Glossiness
- 29. Into “Engine”
- 30. Refining “Quick” in H2:A
- 48. Base Color Pattern Intensity Pattern Tile Pattern Roughness Roughness Intensity Pattern Metallic Metallic Intensity Normal Intensity Cavity Intensity SSS Pattern Adjustment SSS Pattern Intensity
- 51. Refining “Quick” Quick to Automatic: Distilling External and Internal Tools
- 52. ● Studio-wide R&D ● Ancillary tools Distilling Tools
- 53. Utilizing Mozarts Correctly ● Automate process ● Avoid menial process task fatigue ● Avoid rocket powered shit shoveling ● Keep skilled workers doing skilled work ● Allows for tiers of workers
- 54. Mozart Process Perception ● Highly Organized ● Process Driven ● Comfortable tools ● Comfortable process ● Makes sense (to you)
- 55. Mozart Process Reality ● Fatiguing ● Non-interruptable ● Not as fast as you think ● Huge potential for human error ● Only works for your assets
- 56. Tool Automation
- 58. PaperMap Asset Creation Design Production Engine Blockout Massout Concept Assembly “Polish” Ship It
- 59. PaperMap Asset Creation Design Production Engine Blockout Massout Concept Assembly “Polish” Ship It
- 61. High res with material IDs Game res with UVs Material Library Presets Base Texture Maps Game Textures Game File Assembly
- 64. Houdini Poly Reduce 10k 5k 1k
- 65. Exploded Assets
- 66. Auto Compositing
- 67. Baking Maps
- 69. Central Textures
- 70. Master Graph
- 71. Custom Graph
- 77. Sketch
- 78. Sketch Refine
- 79. Form Refine
- 82. Sketch
- 83. Sketch Refine
- 84. Form Refine
- 88. Concept / Asset Creation Concept Production Assembly Engine Auto Mation Informed Iteration Polish
- 89. PaperMap Design Production EngineBlockout Massout Concept/Asset Creation Assembly “Polish” Ship It
- 90. Team Roles
- 92. PaperMap Design Production EngineBlockout Massout Concept/Asset Creation Assembly “Polish” Ship It
- 93. Assembly Auto Mation High Res Source Library Engine Specific Assembly Game Res Roughness Metallic Albedo Cavity Normal Asset Texture Material Engine Pack Combo Flip Normal Y Scene File Directories Concept / Production
- 95. ● “Polish” ● Fix assets ● Remake assets ● Reuse unsuitable assets ● Put out fires ● Memory optimization ● Put out more fires ● …make it look better, time permitting ● Polish ● Run through the game and make the POI’s shine
- 98. Blockout Converter Paper Map “Squishy Assets” Tag Surfaces Attach curves Attach locators Massout Playtest
- 100. Each object is a series of components with parameters
- 101. Each building is a series of components with parameters Cardiology Maternity Ward Neonatal Unit General Surgery Cafeteria Pharmacy Median Patient Income Variations Parameters Median Staff Income Doctors Offices MAKE HOSPITAL
- 102. Each level is a series of components with parameters Range Kill Hotspot Intensity Spread Fun Teabagging Threshold Rage Quit Propagation Noob Accessible Average K/D MAKE LEVEL Random Seed
- 103. Addictive QA Monetization Side Quest Random Nerf RPG Grindey Fantasy Orchestral Repetitive MAKE GAME Fun Side Boob Metacritic 72
- 104. VR Technology The next milestone in entertainment experience
- 105. The Global Virtual Museum ● 1% inventory showroom ● Photogrammetry objects ● Virtual environment swapping ● 99% Inventory Crowd Sourcing
- 106. ●Sporting Events ●Forensic Anthropology ●Forensic Crime Scene Reconstruction ●Cold Case Files ●Off-the-path Tours ●Spelunking ●Genghis Khan ●Show / Movie Tours
- 107. THE END??
- 108. MAKE WORLD PEACE
- 109. MAKE WORLD PEACE Launch Enterprise
- 110. Blurring the Separation Between Concept and World Peace Michael Pavlovich Lieutenant Commander of the Supply Closet
Editor's Notes
- These types of speed things are pretty specific to ZBrush, (beyond the scope of this agnostic pipeline demonstgration) so I'll skip that and just refer them to the online videos for further information; Keep it broad, agnostic, non-ZBrush specific (if you are fast and prefer other programs, by all means, but the basic end requirements –high res source with material IDs—stay the same)
- Start with a rough sketch, just like drawing
- Once you reach a certain point of refinement (may vary by project, artist, specific asset requirements), get it into game The good thing about 3D is that all the time 2D concept artists spend honing their craft with rendering shadows, conveying material properties, lighting, perspective, symmetry, etc… all of that is free in 3D, it’s a very small step with a few techniques that provides them another outlet for the time put in developing design, function, color theory, structure, ideas, etc… Without the time investment in polish-before-evalutation, it’s much easier to kill your babies, break something down, do it right, than “it has to be good enough because we’re out of time” First day concept students can achieve this level of work
- Refining and splitting as you go automatically groups subtools and allows for easy material ID assignment
- Make the process even faster by using / building an Asset Libary (insert mesh brushes, pre-refined models that can retain material ID information)
- Material Assignment and Texture Wear
- Material Assignment and Texture Wear
- Stumble on any of these steps? Can’t see it in engine? Just have presets to roll out? THAT’S OK—no point in continuing to carry buckets of water from the river just the front door canal isn’t finished—it’s still closer, and you can keep iterating and working on it. This leads into “pipeline widening and problem solving” in a nutshell: Find a bottle neck, widen it as much as possible, and iterate and refine tools to keep that water flowing—it may not fix everything, but iteration and tool development from the end of the canal is better than walking all the way back to the river. Iteration Still pretty slow, a lot of “quick” steps, but only one “automatic” step (texturing) that relied on consistent material id’s (not a big deal, but not necessarily ideal depending on the situation) However, many, many questions answered by having a somewhat refined mesh, with materials, in game
- Proving out the substance pipeline with custom materials -PBR game specific shader for more accurate preview -In-game cube maps for accurate HDR reflections, wear and tear, exposing parameters for the team -Custom wear, primary and secondary previews, etc...
- Covenant Materials: proving out custom materials based on previously defined parameters by 343 (in photoshop), extrapolating that info into a node based structure that can have exposed parameters and simplify the process for the eventual end user that bypasses all of the headache and fatigue associated with file management and process
- R&D done by an entire studio of artists that can be implemented studio-wide based on the best results Using the BEST tool for the job, not the most COMFORTABLE tool—One mega Mozart made from 10 chopped up ones! -iteration on unfinished parts add functionality and elegant solutions to finish widening the bottleneck -Ancillary tools (don’t feel stupid about tool requests like the reference viewer) -allowing artist to use tools that best get the job done for them (agnostic across the board as possible)
- -Mozarts frontload process so everybody’s a Mozart (avoid “all the assets this guy does looks awesome, too bad the game looks like shit”) -Avoid who has the best tool to shovel shit faster, instead of eliminating shit shoveling through updated workflows or offloading the legwork to a system. -Tiers of workers: junior guys bouncing around central (keep them well rounded), senior guys specializing in front loading and polish -Avoid the best artist on the team doing several months’ worth of menial tasks that could be relegated to an automated process, that’s a waste of money, and demoralizing, or polish artist’s main job being cleanup
- Game Res: generate from separate objects or “dynamesh”
- Starts with central approved material library, shared by material team through custom filter profiles, previewable using agnostic or game shaders
- Master graph used to house everything a central texture artist needs to distill down to a custom object graph preset
- Custom material selection (sorted by priority), custom wear section (with already set min/max and default presets)
- Exposed custom graph to player, presets already created and materials auto-filled by priority
- Material swapping for the end user if needed
- Removing the bottleneck of asset creation: -reconsider the need for incorrect re-use of
- Removing the bottleneck of asset creation: -reconsider the need for incorrect re-use of assets, need for strip texturing, etc…
- -extrapolate this bottle neck "widening" out to other bottle necks, and eventually prepare for a 1-button solution for a map! -extrapolate each object to be a series of components with parameters (make door)
- -extrapolate each building to be a series of components with parameters (make building)
- -extrapolate each level to be a series of components with parameters (make level) -level based on 2d input -john henry 6 maps, or create and pick the best from a thousand and refine from there?
- -extrapolate design to be a series of components with parameters (make game) -occulus rift to explore areas, experiences, scientific collaboration, massive "correct" worlds procedurally built, based on the core concepts of rapid iteration -john henry 6 maps, or create and pick the best from a thousand and refine from there?
- -extrapolate design to be a series of components with parameters (make game) -occulus rift to explore areas, experiences, scientific collaboration, massive "correct" worlds procedurally built, based on the core concepts of rapid iteration -john henry 6 maps, or create and pick the best from a thousand and refine from there?
- -extrapolate design to be a series of components with parameters (make game) -occulus rift to explore areas, experiences, scientific collaboration, massive "correct" worlds procedurally built, based on the core concepts of rapid iteration -john henry 6 maps, or create and pick the best from a thousand and refine from there?