Hey, I just wanted to say that I'm a huge fan of your style and form of explaining things in Blender and I would definitely be interesting to see some kind of paid extensive course on creating assets or scenes. Have you thought about it? :) Have a nice day!
Great tutorial, just one note though. When channel packing, something I've seen more often is the ARM order; it's simply the order correlation between RGB and the metallic, AO, and roughness. The ease of memorizing the acronym ARM is probably why it's used more often for industry standards. It's often just called an ARM map. ARM->RGB (AO to red | Rough to green | Metallic to blue).
Love it! Great note. As long as the roughness is in the green channel. Apparently green has the best compression and preserves more data for roughness, which is often the more complex map of the three. I love the ARM abbreviation. I'll remember that one!
Just use what satisfies your need. For example, for those assets that I want to have more ganular controls in-engine, I make a specific ACG map (Ambient occlusion, Curvature, Grunge) for procedural stuff.
It can be a real optimization save! Someone mentioned this here already, but I'll make mention again. In the video I did MRO (metallic, roughness, occlusion). It's worth mentioning that what looks to be the more common method is ARM or ORM (occlusion, roughness, metallic). Might be worth mentioning if you're looking to pick up habits that are a bit more common.
It's INSANE how many useful tips you've managed to fit into this one 10 minute video. Genuinely, I've been doing 3D modeling and game development for years now and I haven't been able to learn or figure out most of the stuff in this video.
You can also pack ambient occlusion into the tangent space normal map by simply scaling the normal by AO. You wouldn't even need to unpack it, because any lighting calculations would inherently scale with the dot product of the normal as long as you don't re-normalize it. Some game engines can be tricked by that and effectively give you AO without specifying an AO map or using an AO capable shader at all. Also, for very flat, shiny surfaces (polished coins, mirrors etc.) an 8 bit per channel normal map doesn't cut it because you see banding appear in the normals. In that case, you can just pack nX and nY in a single texture, but use two color channels for each, effectively making it a 16 bit fixedpoint value. The z-coordinate then has to be recovered by sqrt(1 - nX^2 - nY^2) Keep in mind that your baked texture don't all need to be of the same size. While it maybe tempting to bake AO together with metalness/roughness, there are many cases where you need a much higher resolution for metalness than for AO - so it would make sense to factor out the AO map or use the above trick for baking it into the normal map. If your textures are more of a cartoonish flavor where you have lots of sharp stylized features, make sure that all straight lines in your texture run parallel to the image borders in UV space - even if that means a sub-optimal UV unwrap. If any high-contrast lines in texture space run at an angle, you wll see magnification artifacts in form of the well known "vasarely squares". These don't show up if the magnified line is straight in texture space. With clever texture layout, you can completely eliminate those artifacts and get away with a very much reduced texture resolution. Lastly, you may want to check whether your engine supports object space or even world space normal maps.Tangent space normal maps are actually meant to be used for animated objects or in case you need to apply a repeating pattern to a non-planar surface. You don't gain anything with tangent space normals when you're baking them and your object doesn't animate or deform. In those cases, object space normal maps are way faster because they don't need the engine to set up a tangent space frame at every vertex. They also don't suffer from any tangent space related ambiguities like texture space winding order, degenerate UVs or topological singularities.
@@rileyb3d I've followed every step of the map sharpening part (fresh Blender 3.6.2 install) and yet the diffuse map doesnt get affected by that filter node whatsoever I do. Any suggestion? Found the solution: inside the Image Editor I was (wrongly) selecting the baked texture instead of "Render Result" (after that I also connected a Viewer node ||| along with the Composite node ||| to the sharpening filter node)
Thanks for checking it out! I don't know if I can post links here, but if you google "what is channel packing maartenhof" you'll get a really excellent article to read up more on.
Holy moly this is one of the best tutorials I've ever seen. Beautifully clear, not too fast or slow, and exactly what I wanted to know with no filler. Thank you!
One tip, you can denoise a noisy ambient occlusion using the compositor, much like using the diamond sharpen node, just with the denoise node instead. Another thing that I like to do when using the compositor to edit an image is to plug the result to the compositor viewer instead of render output. This way I don't have to set the render resolution the same as the input image, altough it does need a bit of extra step after pressing render, and that is switching the view to viewer node.
This model can be optimized for a game even more, removing all greebles and grids and translating them to baked normal map. It will look just the same as with actual geometry. For open grids alpha channel transparency can be used. Also, baking 2K maps is much faster than 4K.
Fantastic video! This baking info will help a lot. Your approach to showing and not just telling is great. If you were to get a chance to do so, a similar approach to explaining your procedural texturing would be a dream vid for me. It’s a struggle sometimes!
Hey Riley ! Wow ! Thanks for this ! I will definitely consider, buying this course from you, when i want to get serious with modeling myself. Much expertise of yours, going into these 3 Minutes of video! And it's nice and quiet at the same time. No music, that would cover your valuable explanations. Thanks a lot. Greetings from germany, europe, earth 🌏🥰
Would love to add this isn't the only way to optimize for games! There is trim sheets, texture atlases, & tiling textures that can also greatly reduce draw calls and texture memory usage. Those of you wanting to dive deeper should look into those concepts as well! It's counter intuitve but sometimes when building larger scenes its better for an object to have 4 different materials on it (yes that increases draw calls) but its countered by the fact that 50 other objects in the scene also use those same for materials. Thus reducing the total drawcalls for a scene rather then every small object using their own materials / baked textures
The antenna junction above flat surface introduces possible surface Z-fighting. Separate high detail grill insert introduces T-Junctions, so it's best to extend it somewhat so geometry pieces overlap and obscure a crawling dots line, but that too can cause Z-fighting. It depends, sometimes in a game engine for a given game design it may be impossible to set a suitably far nearplane to suppress Z fighting well enough, it would necessitate modelling which isn't prone to these issues, where the topo is cohesive. On the flip side no game engine ever objects against triangulated, decimated models. The quad topology is for your modelling/modification purpose only, and there are topos which are not editor friendly but render absolutely beautifully, think icosphere sections. Oh also long skinny polygons kill ROP bandwidth. If you are triangulating a circle, and really care about its bandwidth consumption (you don't always care), make a big triangle in the middle and arrange smaller ones around it, this way the bulk of the area has no quad overhead. Rendering occurs in 2x2 pixel quads, where whether 4 pixels are occupied in a given triangle or just one pixel, it takes the same time, so it pays to minimise total polygon perimeter and populate the bulk of the area with few polygons.
I'm glad it was helpful! I also love the "simple bake" addon in Blender. It takes the process I just showed, but makes it even faster and more intuitive. For this video I wanted to make sure the foundational principles are there, so I showed it in vanilla Blender.
@@rileyb3dWhich is fantastic because i have been trying to get how this work for years now. Seeing as i ended up just using Substance Painter instead. While i wanted atleast some of the node stuff to also go onto Substance Painter as texture, just never managed to get how baking works.
I was totally unaware that you can use one image, copied over multiple materials, to bake each separate material into a single map. My process would have been to copy the object, create a new material for the new one, and then go for a Selected to Active baking method (which I notice it sucks most of the times).
@@benhunter8551or just use substance painter/designer to bake. They are very cheap. 20 dollars per month and u get both of them with 3d sampler(*it's single photo to pbr material AI is really good specially if u have an nvidia RTX series card.
Just something i wanted to mention regarding the cut tool you used at 0:59 i always use the edge collapse and delete edge loop to terminate loops where possible, there are some clicks to be saved, those work fine instead of the cut tool for the kinda thing you were doing in the vid :)
I would love to see a course or tutorial maybe on how you did the texturing, because i am always interested in how someone would make something like that without using Substance Painter
@@kevlargames1860 and how robust your systems for lighting etc are, since we're taking interactive rather than production class rendering. Because there's a high likelihood that the inside of the TV even if it were fully modelled won't be correctly lit, with some light spilling where it shouldn't and showing wrong directional cues and just looking weird.
@@SianaGearz yeah that’s true too if it where like a product demo then totally have some extra detail but if it’s a real time render then chances are players wouldn’t look that deep into detail so might as just use the black background in most cases
Thanks, as a self punishing perfectionist I wrongly thought a model had to be made from one watertight perfect quad mesh. I should unlearn thinking like that lol.
I'm right there with you! It's taken me a while to accept the reality that I'm better off just disconnecting parts. Once you get the hang of it I think you'll love it. It's a much more forgiving way to go about modeling.
Use the correct words. It's separate meshes not objects...if you separate objects it means more materials, which makes the game laggy. Separate mesh not objects!
@@Dianruakrhakakks I understand what you mean but I was not referring to blender objects, just a 3d mesh in general And you forget, separate objects can all share the same material and UV. (In Blender)
The basic idea is “do it as it is IRL” It's better to have 6 stretched cubes for a simple table than trying to merge legs with frame and top. For the most part - it's both easier and better to make a model out of its parts than it is to model it as one piece. You don't get to mess with topology as much, you can unwrap easier and you can quickly make variations of your model.
@@rileyb3d i liked this video because it showed me a bit of workflow and it looks really profissional . to answer your question i would say i want to see more detailed look into assets pipeline whether its characters, assets or any other field in 3D.- i say this as one who is very decent in modeling and texturing. but in a real work situation i would total be a noob.
great short Tips. thx... renderer Cycle baking settings is also big impotant part and hassle spot ... color space too...special f. beginners... hope you can bring light to that too
Absolutely! I still make bad decisions on where to separate and what to leave unified, but I feel like those are the decisions that improve with practice.
How important is it to create a high poly model first and then baking from that? Especially for the normal maps, that should give better details in the final model
Someone mentioned here already, but I'll bring it up again. In the video I did MRO (metallic, roughness, occlusion, in RGB order). It's worth mentioning that what looks to be the more common method is ARM or ORM (occlusion, roughness, metallic). Might be worth mentioning if you're looking to pick up habits that are a bit more common.
This is good, but also very time consuming. In Substance, i can just paint and export and the softqare does all the manuall stuff for me. I like that this stuff can be done solely in Blender to a VERY high quality.
Thanks for this! Just one question, do we triangulate before unwrapping or just before baking (thinking UE auto lightmaps matching so those verts in such don't change down the line.).
I unwrap before triangulation. When modeling from quads with edge loops, it's much easier to set the seams before triangulation. I haven't personally had issues with this workflow.
You should probably make sure you bake the data textures as linear and not sRGB, they will break or just be generally worse quality in your game engine otherwise.
sRGB is perfectly fine for colour maps, and when saving out other maps like AO, Roughness, or packed ones, etc. You use TGA or similar file format that by default isnt sRGB. Than within the engine you just have to make sure that they do not use sRGB for those maps. Unreal engine detects Normal and packed images on its own
@@sebbosebbo9794 The setting appears both on the node & texture in the image editor. It's not your render color space, just the color space for that texture. You have to rebake for it to take effect.
So you export 4k maps but use 2k for game engines? If so, how does the quality of the image hold up? Does the sharpening tip you shared come through when scaling down the image or does it depend which image format we use (jpeg, jpg, png...) when baking and saving?
2K resolution still generally hold up well, and you can add detail texturing in the game engine. Idea being blending the base textures with scaled textures for the micro details, helping to hide the pixel interpolation upclose.
Notes : 1. Localize geometry : Add edition geometry but only near it. 2. Dissolve the edge that does not affect the silhouette contribution 3. Break up the mesh to not include more geo 4. 1 Single Material less draw call Safe Practice : After triangulating, Texture Baking. Channel Packing = R,G,B Different = Metal, Roughness, Aimbient Occlusion 8:42 Blender Compositor : Color Management standard Compositor Filter node (Diamond Sharpen .5)
Game engines will automatically convert everything into triangles, but automatic conversion may not split a quad the way you would prefer. If you handle the triangulation yourself in Blender, then you can look for and fix any such issues. For example, any quad that isn't perfectly flat will either "bend in" between one pair of opposing vertices or "bend out" between the other pair. This will affect how the quad is shaded, which can lead to visual artifacts when the choice a program makes for the "bend" doesn't match your intended shape. You can even notice this kind of undesired shading artifact when working purely with quads within Blender, as Blender itself has to pick which way to "bend" a quad in order to display it onscreen.
I answered this in another comment, but I'll copy paste it here: In the snapping menu (next to magnet icon on the top of 3D viewport) you can enable "face project". This will snap to the face of an object. You can also hold down shift to combine a few snapping methods. I usually have vertex, edge, and face all selected. Turn on snapping by pressing the magnet icon, or temporarily turn it on by holding Ctrl. I used Ctrl in the video to temporarily snap to the face.
Can someone give me a concrete answer about this: ive been told triangles in modeling and Retopology are very, very bad. Yet i see tons of models online, and sketchfab that use them. NEARLY ALWAYS on hard surface, and sometimes on organic, but near like, the top of the head or places away from the mouth, like inside eye sockets and inside the mouth, places that wont be animated. So my qiestion is: are triangles really bad or not? Is using them a Sin? Will my models be spit upon and i be laughed at for using them?
Models will always be triangulated in game engines. If you model just for still renders, then you can have quads, tris or ngons. Ngons have bad stigma because of shading issues. Think of triangles as a final step. It is pain to UV them, that's why quads are better for that.
A few tips, don't bake your channel packed map, or any map besides base color, as sRGB, you want those to be linear textures, or you will get incorrect results in game engines. Also, why did you bake a normal map without a parent mesh to get normal data from?
In the snapping menu (next to magnet icon on the top of 3D viewport) you can enable "face project". This will snap to the face of an object. You can also hold down shift to combine a few snapping methods. I usually have vertex, edge, and face all selected. Turn on snapping by pressing the magnet icon, or temporarily turn it on by holding Ctrl. I used Ctrl in the video to temporarily snap to the face.
Thanks for having me on this one!
Hey, I just wanted to say that I'm a huge fan of your style and form of explaining things in Blender and I would definitely be interesting to see some kind of paid extensive course on creating assets or scenes. Have you thought about it? :) Have a nice day!
I feel like this should be pinned 🤷
I have to ask do you triangulate only in Blender or you should also do that for Substance Painter?
Great tutorial, just one note though.
When channel packing, something I've seen more often is the ARM order; it's simply the order correlation between RGB and the metallic, AO, and roughness.
The ease of memorizing the acronym ARM is probably why it's used more often for industry standards. It's often just called an ARM map.
ARM->RGB
(AO to red | Rough to green | Metallic to blue).
Love it! Great note. As long as the roughness is in the green channel. Apparently green has the best compression and preserves more data for roughness, which is often the more complex map of the three.
I love the ARM abbreviation. I'll remember that one!
I usually see it abbreviated as ORM in game engines. Godot has a built in material option specifically for handling ORM PBR materials out of the box.
Just use what satisfies your need. For example, for those assets that I want to have more ganular controls in-engine, I make a specific ACG map (Ambient occlusion, Curvature, Grunge) for procedural stuff.
@@ChrisD__ Its exactly the same but use O for "Occlusion" instead of A for 'Ambient occlusion".
ARM!!! thank you. this is easy to remember!
you might've made the most helpful texture baking tutorial I've ever seen for blender. thank you
dude that channel packing tip is genius
It can be a real optimization save! Someone mentioned this here already, but I'll make mention again. In the video I did MRO (metallic, roughness, occlusion).
It's worth mentioning that what looks to be the more common method is ARM or ORM (occlusion, roughness, metallic).
Might be worth mentioning if you're looking to pick up habits that are a bit more common.
It's INSANE how many useful tips you've managed to fit into this one 10 minute video. Genuinely, I've been doing 3D modeling and game development for years now and I haven't been able to learn or figure out most of the stuff in this video.
You can also pack ambient occlusion into the tangent space normal map by simply scaling the normal by AO. You wouldn't even need to unpack it, because any lighting calculations would inherently scale with the dot product of the normal as long as you don't re-normalize it. Some game engines can be tricked by that and effectively give you AO without specifying an AO map or using an AO capable shader at all.
Also, for very flat, shiny surfaces (polished coins, mirrors etc.) an 8 bit per channel normal map doesn't cut it because you see banding appear in the normals. In that case, you can just pack nX and nY in a single texture, but use two color channels for each, effectively making it a 16 bit fixedpoint value. The z-coordinate then has to be recovered by sqrt(1 - nX^2 - nY^2)
Keep in mind that your baked texture don't all need to be of the same size. While it maybe tempting to bake AO together with metalness/roughness, there are many cases where you need a much higher resolution for metalness than for AO - so it would make sense to factor out the AO map or use the above trick for baking it into the normal map.
If your textures are more of a cartoonish flavor where you have lots of sharp stylized features, make sure that all straight lines in your texture run parallel to the image borders in UV space - even if that means a sub-optimal UV unwrap. If any high-contrast lines in texture space run at an angle, you wll see magnification artifacts in form of the well known "vasarely squares".
These don't show up if the magnified line is straight in texture space. With clever texture layout, you can completely eliminate those artifacts and get away with a very much reduced texture resolution.
Lastly, you may want to check whether your engine supports object space or even world space normal maps.Tangent space normal maps are actually meant to be used for animated objects or in case you need to apply a repeating pattern to a non-planar surface. You don't gain anything with tangent space normals when you're baking them and your object doesn't animate or deform. In those cases, object space normal maps are way faster because they don't need the engine to set up a tangent space frame at every vertex. They also don't suffer from any tangent space related ambiguities like texture space winding order, degenerate UVs or topological singularities.
Very helpful, baking is always a pain in the butt. The sharpen trick is really nice too!
Thanks! I didn't have time to cover it in this video, but "simple bake" is a great addon to streamline the baking workflow in Blender.
@@rileyb3d I've followed every step of the map sharpening part (fresh Blender 3.6.2 install) and yet the diffuse map doesnt get affected by that filter node whatsoever I do. Any suggestion?
Found the solution: inside the Image Editor I was (wrongly) selecting the baked texture instead of "Render Result" (after that I also connected a Viewer node ||| along with the Composite node ||| to the sharpening filter node)
Channel Packing is something that always confused me, I could never get my head around it, but this explains the process perfectly, thank you.
Thanks for checking it out! I don't know if I can post links here, but if you google "what is channel packing maartenhof" you'll get a really excellent article to read up more on.
Holy moly this is one of the best tutorials I've ever seen. Beautifully clear, not too fast or slow, and exactly what I wanted to know with no filler. Thank you!
something about a good optimized mesh is just so beautiful to see
One tip, you can denoise a noisy ambient occlusion using the compositor, much like using the diamond sharpen node, just with the denoise node instead.
Another thing that I like to do when using the compositor to edit an image is to plug the result to the compositor viewer instead of render output.
This way I don't have to set the render resolution the same as the input image, altough it does need a bit of extra step after pressing render, and that is switching the view to viewer node.
wow was this one of the best game asset tutorials on the internet. great job.
Thanks! That's great to hear.
finally a gamedev related video, this might be the most useful video you've ever done on this channel.
This model can be optimized for a game even more, removing all greebles and grids and translating them to baked normal map. It will look just the same as with actual geometry. For open grids alpha channel transparency can be used. Also, baking 2K maps is much faster than 4K.
Well this solves the mystery of the channel separation nodes i see in purchased assets. Thank you!
I think this is one of best videos I've seen about baking. Thanks for sharing!
Thanks! Just watched some of your UA-cam stuff as well. Loved the aesthetic of your "moving meditations" submission.
Thanks for watching.
@@rileyb3d Thanks a lot for stopping by, I truly appreciate it 🙏
Fantastic video! This baking info will help a lot. Your approach to showing and not just telling is great. If you were to get a chance to do so, a similar approach to explaining your procedural texturing would be a dream vid for me. It’s a struggle sometimes!
The texture bake explaining was so well and easy thanks alot!
That sharpen trick is really great!
One of my favorites! It might not be appropriate for every style, but I find it especially useful with grungy/realistic assets.
Hey Riley !
Wow ! Thanks for this !
I will definitely consider, buying this course from you, when i want to get serious with modeling myself.
Much expertise of yours, going into these 3 Minutes of video!
And it's nice and quiet at the same time. No music, that would cover your valuable explanations.
Thanks a lot.
Greetings from germany, europe, earth
🌏🥰
golden production level, down to the point, well written, understandable. good
Would love to add this isn't the only way to optimize for games!
There is trim sheets, texture atlases, & tiling textures that can also greatly reduce draw calls and texture memory usage. Those of you wanting to dive deeper should look into those concepts as well!
It's counter intuitve but sometimes when building larger scenes its better for an object to have 4 different materials on it (yes that increases draw calls) but its countered by the fact that 50 other objects in the scene also use those same for materials. Thus reducing the total drawcalls for a scene rather then every small object using their own materials / baked textures
The antenna junction above flat surface introduces possible surface Z-fighting. Separate high detail grill insert introduces T-Junctions, so it's best to extend it somewhat so geometry pieces overlap and obscure a crawling dots line, but that too can cause Z-fighting. It depends, sometimes in a game engine for a given game design it may be impossible to set a suitably far nearplane to suppress Z fighting well enough, it would necessitate modelling which isn't prone to these issues, where the topo is cohesive.
On the flip side no game engine ever objects against triangulated, decimated models. The quad topology is for your modelling/modification purpose only, and there are topos which are not editor friendly but render absolutely beautifully, think icosphere sections.
Oh also long skinny polygons kill ROP bandwidth. If you are triangulating a circle, and really care about its bandwidth consumption (you don't always care), make a big triangle in the middle and arrange smaller ones around it, this way the bulk of the area has no quad overhead. Rendering occurs in 2x2 pixel quads, where whether 4 pixels are occupied in a given triangle or just one pixel, it takes the same time, so it pays to minimise total polygon perimeter and populate the bulk of the area with few polygons.
Will there be a tutorial on this TV? A very interesting part of the procedural texturing of this prop.
Yesss, I can finally use my procedural mats in peace. Thank you!
Finnaly! someone explaining this how to bake a Texture.
I'm glad it was helpful! I also love the "simple bake" addon in Blender. It takes the process I just showed, but makes it even faster and more intuitive.
For this video I wanted to make sure the foundational principles are there, so I showed it in vanilla Blender.
@@rileyb3dWhich is fantastic because i have been trying to get how this work for years now.
Seeing as i ended up just using Substance Painter instead.
While i wanted atleast some of the node stuff to also go onto Substance Painter as texture, just never managed to get how baking works.
Super lucid, expertly delivered - thank you Riley!
Thanks for coming by to watch!
I was totally unaware that you can use one image, copied over multiple materials, to bake each separate material into a single map. My process would have been to copy the object, create a new material for the new one, and then go for a Selected to Active baking method (which I notice it sucks most of the times).
Yes! Valuable feature; however, like a lot of Blender features, it's not extremely obvious.
@@benhunter8551 They could make some UI tweaks to make it more intuitive but I think the process is pretty easy already.
@@benhunter8551 really? nice, where can I find the addon. and is it for free?
@@benhunter8551or just use substance painter/designer to bake.
They are very cheap.
20 dollars per month and u get both of them with 3d sampler(*it's single photo to pbr material AI is really good specially if u have an nvidia RTX series card.
Just something i wanted to mention regarding the cut tool you used at 0:59 i always use the edge collapse and delete edge loop to terminate loops where possible, there are some clicks to be saved, those work fine instead of the cut tool for the kinda thing you were doing in the vid :)
I would love to see a course or tutorial maybe on how you did the texturing, because i am always interested in how someone would make something like that without using Substance Painter
You can also make the vents a 2d plane with a transparent texture like they do with fences
Just make the vents pure absobent black, not even transparent.
@@SianaGearz true! I guess at that point it comes down to how much detail you want
@@kevlargames1860 and how robust your systems for lighting etc are, since we're taking interactive rather than production class rendering. Because there's a high likelihood that the inside of the TV even if it were fully modelled won't be correctly lit, with some light spilling where it shouldn't and showing wrong directional cues and just looking weird.
@@SianaGearz yeah that’s true too if it where like a product demo then totally have some extra detail but if it’s a real time render then chances are players wouldn’t look that deep into detail so might as just use the black background in most cases
Thanks, as a self punishing perfectionist I wrongly thought a model had to be made from one watertight perfect quad mesh.
I should unlearn thinking like that lol.
I'm right there with you! It's taken me a while to accept the reality that I'm better off just disconnecting parts. Once you get the hang of it I think you'll love it. It's a much more forgiving way to go about modeling.
Use the correct words. It's separate meshes not objects...if you separate objects it means more materials, which makes the game laggy. Separate mesh not objects!
@@Dianruakrhakakks
I understand what you mean but I was not referring to blender objects, just a 3d mesh in general
And you forget, separate objects can all share the same material and UV. (In Blender)
The basic idea is “do it as it is IRL”
It's better to have 6 stretched cubes for a simple table than trying to merge legs with frame and top.
For the most part - it's both easier and better to make a model out of its parts than it is to model it as one piece. You don't get to mess with topology as much, you can unwrap easier and you can quickly make variations of your model.
@@zhulikkulik Ye true
Bunch of DOPE tricks. Thanks ❤
The most useful video for 3D artists!
Wow, I've always used the selected to active trick, never knew this way. Subscribed!
Very nice tutorial - thank you very much for sharing :).
Thanks for watching!
amazing tutorials you have shared thanks a lot CG Cookie🙏
Excellent and concise video.
Thanks! I do my best to cut the fluff in my videos. I'm glad it's appreciated.
Loved the sharpening tip, thanks
super useful video! thanks so much
Thanks for taking the time to watch!
Awesome Tips!
Really great tutorial, thanks for sharing!
this is super useful. would love to see more like this
Is there a particular topic you have in mind? Or would you just be happy to see more game dev or asset creation content in general?
@@rileyb3dGame dev asset creation techniques please.
@@rileyb3d i liked this video because it showed me a bit of workflow and it looks really profissional . to answer your question i would say i want to see more detailed look into assets pipeline whether its characters, assets or any other field in 3D.- i say this as one who is very decent in modeling and texturing. but in a real work situation i would total be a noob.
6:25 Here, why do we set it to emit instead of diffuse? Isn't diffuse specifically for the base color?
Finally something on game assets again! Wonderful video :)
It was very helpful, and Baking increases performance.😉
great short Tips. thx...
renderer Cycle baking settings is also big impotant part and hassle spot ...
color space too...special f. beginners...
hope you can bring light to that too
Great video! Thanks.
3:14. Makes total sense. I can work much faster with meshes separated but on the same UV.
Absolutely! I still make bad decisions on where to separate and what to leave unified, but I feel like those are the decisions that improve with practice.
the sharpen part... is so damn smart...
For the vent, would it be better to make an alpha channel with it and just kinda "cut" out the vents and just make it all a texture?
That's great tips... thank you guys !!!
How important is it to create a high poly model first and then baking from that? Especially for the normal maps, that should give better details in the final model
insane cool ! thank you !
Perfect thank you!!
OMG this is insanely good
Thank you, great tips. ❤
Awesome Tut!!! Thanks!
first time hearing about channel pack
Someone mentioned here already, but I'll bring it up again. In the video I did MRO (metallic, roughness, occlusion, in RGB order).
It's worth mentioning that what looks to be the more common method is ARM or ORM (occlusion, roughness, metallic).
Might be worth mentioning if you're looking to pick up habits that are a bit more common.
I request you to make a detailed procedural texturing tutorial of this Television set.
Expecting more videos related to Game Dev.
Thank you!
That model is insanely realistic
thank you for this.
please could you make tutorial on how you textured this model
Wow, so good!!!
Nice tutorial. Just didn't get the point of baking the color as an emission texture instead of a simple diffuse one
This is good, but also very time consuming. In Substance, i can just paint and export and the softqare does all the manuall stuff for me.
I like that this stuff can be done solely in Blender to a VERY high quality.
How to bake displacement textures please can you guide
Good tut!
Please can you cover how you textured this model
please make more game creation using blender....because i am quite interested in this development recently
Thanks for this! Just one question, do we triangulate before unwrapping or just before baking (thinking UE auto lightmaps matching so those verts in such don't change down the line.).
I unwrap before triangulation. When modeling from quads with edge loops, it's much easier to set the seams before triangulation. I haven't personally had issues with this workflow.
Amazing!!!
Nice, there is not enough Blender to game engine tutorials out there !!
Thanks! Hoping to change that.
Doing it this way doesn't require you unplugging metallic node. Cool.
Yeah, since we're isolating the diffuse (as far as I know), metallic will not interfere.
What if you have to turn a high-poly mesh to a low-poly one?
Beethoven, 7th Symphony, 2nd Movement at 00:42 ? Good video ;)
You should probably make sure you bake the data textures as linear and not sRGB, they will break or just be generally worse quality in your game engine otherwise.
is this a node setting... f.e. normal baking change this setting even too non linear..? or Im wrong...is it a color space or render setting change ..!
sRGB is perfectly fine for colour maps, and when saving out other maps like AO, Roughness, or packed ones, etc. You use TGA or similar file format that by default isnt sRGB. Than within the engine you just have to make sure that they do not use sRGB for those maps. Unreal engine detects Normal and packed images on its own
@@SolarTaraYeah I forgot to specify that I was talking about the textures with just data in them.
@@sebbosebbo9794 The setting appears both on the node & texture in the image editor. It's not your render color space, just the color space for that texture. You have to rebake for it to take effect.
So you export 4k maps but use 2k for game engines? If so, how does the quality of the image hold up? Does the sharpening tip you shared come through when scaling down the image or does it depend which image format we use (jpeg, jpg, png...) when baking and saving?
2K resolution still generally hold up well, and you can add detail texturing in the game engine. Idea being blending the base textures with scaled textures for the micro details, helping to hide the pixel interpolation upclose.
Are the topology changes necessary if you are using nanite in ue5?
Notes :
1. Localize geometry :
Add edition geometry but only near it.
2. Dissolve the edge that does not affect the silhouette contribution
3. Break up the mesh to not include more geo
4. 1 Single Material less draw call
Safe Practice :
After triangulating, Texture Baking.
Channel Packing = R,G,B Different = Metal, Roughness, Aimbient Occlusion
8:42
Blender Compositor :
Color Management
standard
Compositor
Filter node
(Diamond Sharpen .5)
Hey loved the video, mind sharing the texture file?
strange music at the end... oddly inspiring
Grills also could be bake too
But why triangulate? Is there some benefit from keeping them all quads ?
Game engines will automatically convert everything into triangles, but automatic conversion may not split a quad the way you would prefer. If you handle the triangulation yourself in Blender, then you can look for and fix any such issues.
For example, any quad that isn't perfectly flat will either "bend in" between one pair of opposing vertices or "bend out" between the other pair. This will affect how the quad is shaded, which can lead to visual artifacts when the choice a program makes for the "bend" doesn't match your intended shape. You can even notice this kind of undesired shading artifact when working purely with quads within Blender, as Blender itself has to pick which way to "bend" a quad in order to display it onscreen.
Nice!
What is the music in the end of the video?
Great tutorial, Thank you.
and how do i use the channel pack in the game engine?
Amazing tutorial -
But for the love of all that is holy, WHY THE MUSIC????
( still subscribed fyi )
its sound or feels like I watch the end of the last tutorial of all time before😅😂🤣💣💥🗯💭💨🕊 a dramatic super Gau happens.
2:12 wow, how did you move that so precisely and fast?
I answered this in another comment, but I'll copy paste it here:
In the snapping menu (next to magnet icon on the top of 3D viewport) you can enable "face project". This will snap to the face of an object. You can also hold down shift to combine a few snapping methods. I usually have vertex, edge, and face all selected.
Turn on snapping by pressing the magnet icon, or temporarily turn it on by holding Ctrl. I used Ctrl in the video to temporarily snap to the face.
oh, thank you very much dear sir@@rileyb3d
Can someone give me a concrete answer about this: ive been told triangles in modeling and Retopology are very, very bad. Yet i see tons of models online, and sketchfab that use them. NEARLY ALWAYS on hard surface, and sometimes on organic, but near like, the top of the head or places away from the mouth, like inside eye sockets and inside the mouth, places that wont be animated. So my qiestion is: are triangles really bad or not? Is using them a Sin? Will my models be spit upon and i be laughed at for using them?
Models will always be triangulated in game engines. If you model just for still renders, then you can have quads, tris or ngons. Ngons have bad stigma because of shading issues. Think of triangles as a final step. It is pain to UV them, that's why quads are better for that.
Sharpening: Certainly a brilliant idea, but: How do you make the image appear - when I drop an image into the compositor: Nothing happens - what now?
A few tips, don't bake your channel packed map, or any map besides base color, as sRGB, you want those to be linear textures, or you will get incorrect results in game engines. Also, why did you bake a normal map without a parent mesh to get normal data from?
It's Awesome
how did you slide that antenna ontop the surface that fast?
I'm doing it manually and it's not perfect even.
surface snap
In the snapping menu (next to magnet icon on the top of 3D viewport) you can enable "face project". This will snap to the face of an object. You can also hold down shift to combine a few snapping methods. I usually have vertex, edge, and face all selected.
Turn on snapping by pressing the magnet icon, or temporarily turn it on by holding Ctrl. I used Ctrl in the video to temporarily snap to the face.
Ничего не понял, но очень интересно!
I'd like to know the name of the music used in the video.
"Inevitable" by Risian
@@rileyb3d thanks
Tooop....Thank you!!!