Sign up for my upcoming 3D Modeling course today: www.onmars3d.com/ And my Patreon: www.patreon.com/OnMars3D for all my project files, 3d models, bonus videos, Discord and more or just to support this channel 🙂 Discord: discord.gg/vybvJnEeya
After taking a 5+ year break from 3D modelling to look after my mom I'm so happy to see these principals still in place and taught by lovely people like yourself. Keep up the great work and great to see how far you've come Mars!
@@JamesSmith-ix5jdI came here for an answer to this question as well and it seems no one has an answer, at least, not in the last 12 days haha. Maybe everyone else already knows how to add details to the subD mesh? My guess is to apply SubD in Blender and then add details, but then you can't go back to low poly... Idk
I'm not much of a commenter in general, but the sheer quality and value of your content is absolutely incredible. The approach is always clear and the workflows are just excellent all round. Thankyou for your effort, sincerity and passion.
You definitely give enough information on each topic where it fufills any concerns without going into unecessary explaining or bringing in other side topics to confuse with!!@@OnMars3D
What you said in the video is basically what my 3D modeling instructor said this morning. I'm so glad newbies like me have people like you or my instructor.
Agreed, especially on the last tip. That is what is used in the AAA car models but you rarely see anyone on YT teaching that way and as a result, their models always have shading errors.
I'm learning hard-surface and SubD modeling at the moment, and that last tip was huge. I kept beating my head against a wall trying to add details to my unsubdivided mesh, which was leading to tons of later shading issues and unnecessary topology. Working to the main form, then working smaller is a game changer. Thank you so much!
The last tip is something I use frequently when I 3D model cars, where clean reflections are paramount. Making sure all the large surfaces are perfect before considering fine details - and then post smoothing you work manually on those details. This is why it's important to not just learn high-poly or low-poly. You can benefit greatly from practicing both. They compliment each other, and understanding polygonal detail work through low-poly workflow is highly beneficial for high-poly projects when it's time to add the fine details without relying on smoothing them artificially.
Yeah, I have the same reaction as you did when I discovered (with you) about the details over the subdivided mesh, and not the base mesh. I always did like you did before: trying hard to put everything in the base mesh, trying to do in the most clean and low poly way as possible. This gave me headaches in some models, Jesus! I discovered about this watching your Datsun videos that you showed now in this video (trying to model a Miata Mx5, and doing well untill now). If I get enough time to end the model, I'll try to show to you, because your videos was part of the entire process. I work in 3D industry for a decade, but in Archviz area. Usually, we don't need to model too much, we model simple things. The more complex models are specifc furnitures, but it's way easier than a car, for example. Furniture is usually handmade, and it's not very precise, sometimes we model without blueprints with just one terrible perspective photo, it's small. But with car, I can step up my modelling skill. Try to get precise highlights that goes through the entire car body is way harder. But it was a happy moment when I achieve a great result, like all the references that I got. Anyway. Great content, your channel is awesome!
@@The_Curious_Ra.ke01after adding the details what next?? Agn applying subd right? If we don't apply the subd after detailing then the details we added later in the subd mesh there border n edges will remain hard and not smooth.. So i m asking.. Kindly answer it will help...
Rare practical advice aimed at serious 3d modelers. I very much like the context you give this video and plan to dive into your content. Thank you for sharing.
Because of that Artstation post, I paused and spent an hour or so to figure out how to get a clean surface topo similar to his first example. While I did eventually realize the correct way as how the senior modeler explained, I wish I simply watched the whole video instead. The devil truly is in the details.💀-1hr Also, it's crazy how you're way on point with these videos! Sets this channel apart from the rest.
the way how i deal with tri and Ngon is treat it like art picture if it isn't in the focus view don't put too much paint on it, if the tri or Ngon in an area that isn't affect by the bone riging and edge leave it be could save your sanity and polycount
Last advice is really eye opening. I never thought to add details after subdivision. I always try hard to put all details in base mess and then apply subdivision.
My thoughts during tips 1-4 "Ok, ok, i know that, ok". listens to tip number 5 *Mind blown*. This is why I love watching tips and trick videos, there is always something new to learn:D
Hey Aunmar, I just wanted to thank you for your educative content that helped me land my first internship in the game industry as a vehicle artist. Your Datsun series taught me a lot. Thank you
Good video! Some extra for the tip at 8:02 though: Consider using Crease on the edges that you want to stay hard before subdividing, it prevents them from smoothing so you can still easily add the support loops around them
the beautiful thing about proper sub-d modeling workflow is you can make a low poly model, that is also a hero quality film ready model by just adding edge loops in the right way
That last tip was a realization I had about a year ago while also working on car model. The only downside is that you better be pretty certain you don’t need to make any super low frequency changes to the shape, or you may end up doing retopo sooner than you planned 😅 Solid tips! I’m definitely going to check out the artists mentioned in the video.
0:38 This where those misnamed “poles” (actually “extraordinary vertices”) come in. In order to have fewer edge lines in some places and more in others, you need to be able to branch them where they join. 5-verts and 3-verts are useful for this purpose.
We need content specifically about number four. Many graphic designers familiar with bezier curves instinctively try to use side effects of subdivision surface to control form with as little detail as possible. When we learn that for subdivs to be used properly we need 6x or 8x more geometry before adding subdivs, we get stuck. We don't know how to work with that much geometry manually without it taking dozens of hours that we don't have.
Great tips. I do stuff mostly for VR in game engines, so render budgets are tight. The last tip is huge. It saves time and leaves your options open for planning the finer detail stuff later. Not every detail needs to be in the mesh, just the defining and prominent lines and silhouettes/shadow casters to support the shape stably under lighting and animation. If it's some deferred or still image render or ultra close-up with intricate details with advanced lighting or needs to still look sharp on raytraced water reflections use as many polys as it needs and never complain about the extra render time lol. Things like bump and reflect map shaders can make a low poly form look more detailed than it really is sometimes but those have budgets too. A master texture artist can also augment that even further with their use of shadow/highlights on the base texture to support the shaders or not even need shaders for some things. You can spread the detail work out to get the best use of render budget and vRAM depending on the goal of the project.
Thank you for your time making this video, I have started to journey of 3D modeling and felt pretty much stuck with the exact issues you have presented. Especially the fifth, no sub-d application until the model is done.
1:09 I would add one or two inset edge loops around that triangle fan. That way, when you add a subdivision surface modifier, you won’t get that weird wrinkly effect.
To follow along on that last tip, if you are making a real time 3d model/game asset you can also bake many of the details from the high poly into a low poly mesh, and you could be very surprised how much you can get away with it.
My biggest problem when learning blender and Poly count is that I was making skins for a game (not at an employee, just for fun). But I have no idea what the ideal polycount should be and I kinda hoped to have a more standardized answer as I really was hoping for the model to be added to the game as they do sometimes (Steam community workshop). I decided to take some models from the game and use their polycount as a guideline. By doing that I was able to know what to aim for.
Low-Poly Base for Form [Maya/Blender] -> Sub-D Higher Detail Mesh [Maya/Blender] -> Low Poly Optimized to fit Sub-D High Poly [Maya/Blender] -> Unwrap Optimized Low-Poly Base [Maya/Blender] -> Bake maps (Normal/AO/ORM/etc.) from HighPoly to Low-Poly Base using [XNormal/SubstancePainter] -> Apply intricate details, textures and materials [Photoshop/Krita/SubstancePainter] -> Smile :)
I used to model in Fusion360, export to 3dsMax and retopo and then create the high poly. Now I Model in a low Fusion360, copy the mesh and add all the high poly details. Export as a low detail obj , export the high as a high detail obj to bake with. The result is a triangulated low , a slight pain to UV but it has a very low vert count and high accuracy in terms of form. This workflow is also very fast.
Great tips. I learned all these through trial and error except that last one. I would really love to see that last tip in action. Maybe in a UA-cam short 😁
Tri count and tri usage are definitely ones I struggle with. Also, the tip about protecting the extrusion I just did for a Sculptober piece. Not even sure why I did it but it just felt right to do. Unfortunately the tip about detailing after SubD I failed to do on that very same piece so there is a small error when it comes to a hole in a cylinder but the silhouette is fine so it worked out lol
For game models you will also have to care about your uv splits, hard edges and normals in general. uv splits and hard edges will increase your vertex count and often times you will need to use face weighted normals with simple bevels to create a nice high poly like shading. But even when you're baking from highpoly, the normals in your lowpoly still matter a lot and will determine if your normal maps will give a good result and compress well. It's also important to know that very feint normal map details will quickly get crushed by texture compression and mipmapping, so sometimes it's better to exaggerate the bevels a little.
Also when thinking about gpu performance in regards to polycount, you should always have in mind that gpus work in blocks of 2x2 pixels generally and if they have to shade a triangle that is only one pixel small, the gpu will do the work for 3 additional pixels for nothing. So the size of the model on the screen and the target resolution are defining factors in how polycount affects performance (to a degree obviously). That's one of the big issues that Unreal Engine 5s Nanite System solves.
In the past 10 years I have watched countless videos about topology, but I still can't create anything because of that. I wish I had an instructor or a friend that could help me
You should join my Discord: discord.com/invite/83MMpPrE We've got a bunch of people helping each other out from modeling and topology to texturing and rendering.
Great video and excellent points. I can see it takes time and effort to put something like this together. So, congrats! Thanks for the mention also! ☺🤙
I have a part on my model that I couldn't figure out how to get rid of a tri, but I decided to leave it there because it's in the middle of my character's horns, it's very small, no one would see it and it will never move
Iv always added support loops around my extrusions. I didn’t know that’s a legit technique, this whole time I thought I was cheating/ doing it improperly
In my self taught adventure to learn 3D modeling, I usually run into the issue with poles in tutorial videos. The #1 thing I hear is to avoid them unless necessary. When you're just learning, you have no idea what is a necessary pole or not. I saw surprised there wasn't anything mentioned here. It makes me think that at a professional level, it is obvious that poles are abundant and necessary to do any high quality modeling.
I don’t do any soft body stuff but I know the poles found on faces (actual faces for characters), specifically the location have been found to be the best over time. I do car modeling related stuff, and every front fender has a strategically placed pole that connects the circular, wheel part, to the flat top where the hood connects. It’s definitely advanced stuff that I wish to learn more about. Same boat, self taught adventures :)
That last tip hit me the same when I found out about it while watching some tutorial. I was also like, wait we can do this? Why didnt I do this earlier and I felt like a complete dummy :D Great tips otherwise, really nicely put video
Well, this video makes me feel good about myself. I've been learning how to model and retop, and it's been a lot of trial and error, lots of restarts, and I've definitely made just about every single one of these mistakes, but I've been making them less and less and less 😭 Thank you for this video, and def will check out those game artists.
Lower polycount means less surfaces to calculate when doing collision detection. Good developers will use generic shapes instead of the base mesh, for obvious reasons, but if you want to do accurate surface detection, say for decal painting, you're pretty much forced to use a low poly solution. If not, well, you'd better have a damn good at spatial partitioning algorithm up your sleeve.
Another quick note, if you make assets for games, game engines render and calculate models in tri’s anyways. Still want good tapo, but you’re not going to get fried if you have a tri here and there.
Also trying to get pockets and zipper flaps into pants is not only something I use to struggle with but still do. The problem isnt getting them to look right tho, its getting it properly rigged and weight painted.
i struggle soo hard with topology i can sculpt and texture no problem but man topology kills me sooo badly, It takes me out of the work creatively and grinds my production to a halt.
question for add detail in subdivided level. it's depend on the context right?, and for that case, it's for the model meant to be baked if it's not the case, the lowest subdiv still need to capture the silhouette right?
hello and thank you for this great tutorial. i have a question : if i use quad topology and triangle topology in my 3D model, is this wrong? my 3D models are rims for marketplace and when I use the bevel modifier and select the arc option from the geometry section and miter outer section, the topology becomes n-gon and quad. that I have to triangulate them in order to correct them.
I am still having issues when it comes to the wavy pattern on cylinders when baking, most of the time 32 edges is enough to prevent that, but when i want to go lower like 24 or less edges, it just wont work. I have read the solutions to this online of baking with more edges and then removing edges after and that has worked in the past but for some details that more complex it just isn't feasible. If you made a video on baking cylinders, that would be a lifesaver for a self-learning guy like me wanting to model environment art for games.
You’re talking about the triangle fans on the end faces? The simple fix is to add one or two supporting loops close to the circumference. That will fix the wrinkly-subsurf problem.
@@OnMars3D yeah, also the super black highlights or the opposite effect of white highlights when trying to bake a high to lowpoly soda can around the rim for example is something I’m trying to figure out. For example the new Spider-Man game has a lowpoly soda can and I wonder if they even used a high poly because their can is like 16 sides but has no baking artifacts from what I can tell.
Woaw amazing for all that, I'm a junior that just started in the vfx/animation industry and all those tips are gold mine for me, I see myself so much in your last tip ahah Thanks !
3:36 I was looking at these images and what I find strange is that the details that are additive are not part of the base mesh, meanwhile the negative details are. But in the subd2 mesh, all the details appear to be part of the base mesh including the additive details. Almost like they are combined with a boolean then beveled. How was this done?
if you look more closely, the details on subd2 mesh are indeed separate, you can actually boolean them, but it would take more time and skills to pull off. As to why they look connected, I think it's just clever trick of the light, and maybe ambient occlusion. There is a trick that I normally use to fake a connection, which is to extrude out the edges so that it barely touches a surface underneath, if that make any sense
MY issue is I still haven't gotten the whole support loops and redirect edge flow down. still tryna undeerstand it and know when to add and when not to add
5:04 I'm very interested in this. Do you have video on this? Also, my only criticism for your video is, you going SUPER FAST over important things. Would've been nice if you slowed down and present the tip by simple/quick demonstration instead of listing a grocery list.
Sign up for my upcoming 3D Modeling course today: www.onmars3d.com/
And my Patreon: www.patreon.com/OnMars3D for all my project files, 3d models, bonus videos, Discord and more or just to support this channel 🙂
Discord: discord.gg/vybvJnEeya
After taking a 5+ year break from 3D modelling to look after my mom I'm so happy to see these principals still in place and taught by lovely people like yourself. Keep up the great work and great to see how far you've come Mars!
Thank you, it's been a fun process!
Can somebody explain the last tip?? U subd first - apply it, then add geometry and subd again?
I love the 'base -> subdiv -> details' tip. Makes so much sense!
Glad it helped!
really helpful advice!
How do you add details to the subdivided surface? By dragging faces around? Or selection masks with falloff?
@@JamesSmith-ix5jdI came here for an answer to this question as well and it seems no one has an answer, at least, not in the last 12 days haha. Maybe everyone else already knows how to add details to the subD mesh?
My guess is to apply SubD in Blender and then add details, but then you can't go back to low poly... Idk
@@JamesSmith-ix5jd same shit. Why they dont unswer?))
I'm not much of a commenter in general, but the sheer quality and value of your content is absolutely incredible. The approach is always clear and the workflows are just excellent all round. Thankyou for your effort, sincerity and passion.
Appreciate the kind words, thank you!
You definitely give enough information on each topic where it fufills any concerns without going into unecessary explaining or bringing in other side topics to confuse with!!@@OnMars3D
What you said in the video is basically what my 3D modeling instructor said this morning. I'm so glad newbies like me have people like you or my instructor.
Agreed, especially on the last tip. That is what is used in the AAA car models but you rarely see anyone on YT teaching that way and as a result, their models always have shading errors.
I'm learning hard-surface and SubD modeling at the moment, and that last tip was huge. I kept beating my head against a wall trying to add details to my unsubdivided mesh, which was leading to tons of later shading issues and unnecessary topology. Working to the main form, then working smaller is a game changer. Thank you so much!
The last tip is something I use frequently when I 3D model cars, where clean reflections are paramount. Making sure all the large surfaces are perfect before considering fine details - and then post smoothing you work manually on those details.
This is why it's important to not just learn high-poly or low-poly. You can benefit greatly from practicing both. They compliment each other, and understanding polygonal detail work through low-poly workflow is highly beneficial for high-poly projects when it's time to add the fine details without relying on smoothing them artificially.
Why wasn’t I taught these in my classes? Welp anyway at least the world has people like you
Appreciate it!
The tip number 5 just changed my life. OMG man that's really amazing
Yeah, I have the same reaction as you did when I discovered (with you) about the details over the subdivided mesh, and not the base mesh.
I always did like you did before: trying hard to put everything in the base mesh, trying to do in the most clean and low poly way as possible. This gave me headaches in some models, Jesus!
I discovered about this watching your Datsun videos that you showed now in this video (trying to model a Miata Mx5, and doing well untill now). If I get enough time to end the model, I'll try to show to you, because your videos was part of the entire process.
I work in 3D industry for a decade, but in Archviz area. Usually, we don't need to model too much, we model simple things. The more complex models are specifc furnitures, but it's way easier than a car, for example. Furniture is usually handmade, and it's not very precise, sometimes we model without blueprints with just one terrible perspective photo, it's small. But with car, I can step up my modelling skill. Try to get precise highlights that goes through the entire car body is way harder. But it was a happy moment when I achieve a great result, like all the references that I got. Anyway.
Great content, your channel is awesome!
Thank you, much appreciated!
I also had same reaction. I also try to put all details in base mesh mess and then apply subdivision. But final advice is really eye opening.
So is the subd applied to the object then you go back into to add the details?
@@The_Curious_Ra.ke01after adding the details what next?? Agn applying subd right? If we don't apply the subd after detailing then the details we added later in the subd mesh there border n edges will remain hard and not smooth.. So i m asking.. Kindly answer it will help...
Rare practical advice aimed at serious 3d modelers. I very much like the context you give this video and plan to dive into your content. Thank you for sharing.
That last tip!!! SO simple yet never passed my mind, thanks man good video
Because of that Artstation post, I paused and spent an hour or so to figure out how to get a clean surface topo similar to his first example. While I did eventually realize the correct way as how the senior modeler explained, I wish I simply watched the whole video instead. The devil truly is in the details.💀-1hr
Also, it's crazy how you're way on point with these videos! Sets this channel apart from the rest.
Glad to hear it, and thanks for the kind words!
that last tip is a game changer for me too.
the way how i deal with tri and Ngon is treat it like art picture if it isn't in the focus view don't put too much paint on it, if the tri or Ngon in an area that isn't affect by the bone riging and edge leave it be could save your sanity and polycount
That last tip was super key for me too. Not many people talk about it
Last advice is really eye opening. I never thought to add details after subdivision. I always try hard to put all details in base mess and then apply subdivision.
man i loved your last tip
it took 2 years for me to learn that LOL. your amazing
Glad it helped my friend!
Last tip is really useful. Thank you
Thanks for making sense of 3d modelling😊
You just blew my mind with that last tip. Jeeze. I feel like i should've realized that
sir, you are great inspiration
Many many thanks
That liast tip and support loops are really important and I do forget.
My thoughts during tips 1-4 "Ok, ok, i know that, ok". listens to tip number 5 *Mind blown*. This is why I love watching tips and trick videos, there is always something new to learn:D
Hey Aunmar, I just wanted to thank you for your educative content that helped me land my first internship in the game industry as a vehicle artist. Your Datsun series taught me a lot. Thank you
That's really awesome, glad to help!!
Good video! Some extra for the tip at 8:02 though: Consider using Crease on the edges that you want to stay hard before subdividing, it prevents them from smoothing so you can still easily add the support loops around them
💯🎯
Thanks for the tip, hoping to cover creasing in future videos!
That's content for me! I feel like the algorithm listened to my thoughts. It's scary, but I'm happy 😅
It's scary how good the algorithm can be 😂
the beautiful thing about proper sub-d modeling workflow is you can make a low poly model, that is also a hero quality film ready model by just adding edge loops in the right way
Tip 5 is gold. As a VERY new modeler (
That last tip was a realization I had about a year ago while also working on car model. The only downside is that you better be pretty certain you don’t need to make any super low frequency changes to the shape, or you may end up doing retopo sooner than you planned 😅
Solid tips! I’m definitely going to check out the artists mentioned in the video.
0:38 This where those misnamed “poles” (actually “extraordinary vertices”) come in. In order to have fewer edge lines in some places and more in others, you need to be able to branch them where they join. 5-verts and 3-verts are useful for this purpose.
We need content specifically about number four. Many graphic designers familiar with bezier curves instinctively try to use side effects of subdivision surface to control form with as little detail as possible. When we learn that for subdivs to be used properly we need 6x or 8x more geometry before adding subdivs, we get stuck. We don't know how to work with that much geometry manually without it taking dozens of hours that we don't have.
Great tips. I do stuff mostly for VR in game engines, so render budgets are tight. The last tip is huge. It saves time and leaves your options open for planning the finer detail stuff later. Not every detail needs to be in the mesh, just the defining and prominent lines and silhouettes/shadow casters to support the shape stably under lighting and animation. If it's some deferred or still image render or ultra close-up with intricate details with advanced lighting or needs to still look sharp on raytraced water reflections use as many polys as it needs and never complain about the extra render time lol. Things like bump and reflect map shaders can make a low poly form look more detailed than it really is sometimes but those have budgets too. A master texture artist can also augment that even further with their use of shadow/highlights on the base texture to support the shaders or not even need shaders for some things. You can spread the detail work out to get the best use of render budget and vRAM depending on the goal of the project.
Great points, thanks for sharing!
Thank you for your time making this video, I have started to journey of 3D modeling and felt pretty much stuck with the exact issues you have presented. Especially the fifth, no sub-d application until the model is done.
I'm actually at your tip 5 stage! Thanks for mentioning that
1:09 I would add one or two inset edge loops around that triangle fan. That way, when you add a subdivision surface modifier, you won’t get that weird wrinkly effect.
Great video
To follow along on that last tip, if you are making a real time 3d model/game asset you can also bake many of the details from the high poly into a low poly mesh, and you could be very surprised how much you can get away with it.
Yup, I talk about baking high to low earlier in the video here: ua-cam.com/video/wRPtxydqDuE/v-deo.html
The idea of of adding details after you subdivide is breaking my brain right now haha but I’m interested to hopefully try it out in the future!
Give it a shot and let me know how it works out in the Discord!
My biggest problem when learning blender and Poly count is that I was making skins for a game (not at an employee, just for fun). But I have no idea what the ideal polycount should be and I kinda hoped to have a more standardized answer as I really was hoping for the model to be added to the game as they do sometimes (Steam community workshop). I decided to take some models from the game and use their polycount as a guideline. By doing that I was able to know what to aim for.
woah! really? the last one was something amazing. I'm glad I watched the full video. Thank you man
Low-Poly Base for Form [Maya/Blender] -> Sub-D Higher Detail Mesh [Maya/Blender] -> Low Poly Optimized to fit Sub-D High Poly [Maya/Blender] -> Unwrap Optimized Low-Poly Base [Maya/Blender] -> Bake maps (Normal/AO/ORM/etc.) from HighPoly to Low-Poly Base using [XNormal/SubstancePainter] -> Apply intricate details, textures and materials [Photoshop/Krita/SubstancePainter] -> Smile :)
I used to model in Fusion360, export to 3dsMax and retopo and then create the high poly. Now I Model in a low Fusion360, copy the mesh and add all the high poly details. Export as a low detail obj , export the high as a high detail obj to bake with. The result is a triangulated low , a slight pain to UV but it has a very low vert count and high accuracy in terms of form. This workflow is also very fast.
Great tips. I learned all these through trial and error except that last one. I would really love to see that last tip in action. Maybe in a UA-cam short 😁
great tips, that first issue is a problem, gonna watch your tut and see if it helps avoiding whole model edge loops :D
Thanks a lot! It's look like simple issues, but so difficult to figured out for myself. Subscribed.
Thanks for the tips helping out beginners!
Thanks for the video! It’s super helpful, the mistake 5 blew my mind as well 😮😮 I’ll definitely change the way I do subd modeling
Glad to hear that!
Wow, that was really useful. Thanks a lot!
This is a good refresher sir thank you!
Glad it was helpful!
👏👏👏Great Video! Essential and straight to the point!
Always a great teacher, you've come a long way since IADT!
I appreciate that!
Looks like a very useful channel to follow.
Glad you think so!
Tri count and tri usage are definitely ones I struggle with. Also, the tip about protecting the extrusion I just did for a Sculptober piece. Not even sure why I did it but it just felt right to do. Unfortunately the tip about detailing after SubD I failed to do on that very same piece so there is a small error when it comes to a hole in a cylinder but the silhouette is fine so it worked out lol
All good, it's part of the learning process!
Hey dude. I learned lot of modeling techniques from you. Practical explanation. Please make more content for us
Glad to hear it, more to come!!
Great tutorial! One suggestion- please use higher contrast colors for your example shots. Black lines on blue is hard to see 😊
Noted, thanks for the suggestion!
For game models you will also have to care about your uv splits, hard edges and normals in general. uv splits and hard edges will increase your vertex count and often times you will need to use face weighted normals with simple bevels to create a nice high poly like shading. But even when you're baking from highpoly, the normals in your lowpoly still matter a lot and will determine if your normal maps will give a good result and compress well. It's also important to know that very feint normal map details will quickly get crushed by texture compression and mipmapping, so sometimes it's better to exaggerate the bevels a little.
Also when thinking about gpu performance in regards to polycount, you should always have in mind that gpus work in blocks of 2x2 pixels generally and if they have to shade a triangle that is only one pixel small, the gpu will do the work for 3 additional pixels for nothing. So the size of the model on the screen and the target resolution are defining factors in how polycount affects performance (to a degree obviously). That's one of the big issues that Unreal Engine 5s Nanite System solves.
In the past 10 years I have watched countless videos about topology, but I still can't create anything because of that. I wish I had an instructor or a friend that could help me
You should join my Discord: discord.com/invite/83MMpPrE
We've got a bunch of people helping each other out from modeling and topology to texturing and rendering.
@@OnMars3D That is very kind I appreciate a lot !!
Great video and excellent points. I can see it takes time and effort to put something like this together. So, congrats!
Thanks for the mention also! ☺🤙
Thanks Bruno, appreciate your models as well!
I don't quite understand the last tip about adding detail after subd and how that would help, any vids on the subject I can check out?
very informative and useful
Glad it was helpful!
Thank you very much for your videos!
Phenomenal advice, 100%
Thank you, sir.
You bet!
Great video my man
Appreciate it
Excellent advice
Awesome video man!
I have a part on my model that I couldn't figure out how to get rid of a tri, but I decided to leave it there because it's in the middle of my character's horns, it's very small, no one would see it and it will never move
Still dont understand the "use a subd and and details" part, would love a video on it!
I've gotten this question quite a bit, I'll work on this soon!
Iv always added support loops around my extrusions. I didn’t know that’s a legit technique, this whole time I thought I was cheating/ doing it improperly
Nope, you're doing it right!
In my self taught adventure to learn 3D modeling, I usually run into the issue with poles in tutorial videos. The #1 thing I hear is to avoid them unless necessary. When you're just learning, you have no idea what is a necessary pole or not. I saw surprised there wasn't anything mentioned here. It makes me think that at a professional level, it is obvious that poles are abundant and necessary to do any high quality modeling.
I don’t do any soft body stuff but I know the poles found on faces (actual faces for characters), specifically the location have been found to be the best over time. I do car modeling related stuff, and every front fender has a strategically placed pole that connects the circular, wheel part, to the flat top where the hood connects. It’s definitely advanced stuff that I wish to learn more about. Same boat, self taught adventures :)
Poles are unavoidable but something you want to use to control the edge flow. Great point!
That last tip hit me the same when I found out about it while watching some tutorial. I was also like, wait we can do this? Why didnt I do this earlier and I felt like a complete dummy :D
Great tips otherwise, really nicely put video
Definitely felt the same 😂
Well, this video makes me feel good about myself. I've been learning how to model and retop, and it's been a lot of trial and error, lots of restarts, and I've definitely made just about every single one of these mistakes, but I've been making them less and less and less 😭
Thank you for this video, and def will check out those game artists.
It's definitely a process, glad you enjoyed it!
thank you big time!!
Lower polycount means less surfaces to calculate when doing collision detection. Good developers will use generic shapes instead of the base mesh, for obvious reasons, but if you want to do accurate surface detection, say for decal painting, you're pretty much forced to use a low poly solution. If not, well, you'd better have a damn good at spatial partitioning algorithm up your sleeve.
Very helpful video, even as a newbie. And I'm guessing knowing when to subd before detailing is mostly a matter of experience and practice, right?
Another quick note, if you make assets for games, game engines render and calculate models in tri’s anyways. Still want good tapo, but you’re not going to get fried if you have a tri here and there.
Really helpfull, cheers.
Glad it helped
I am new to modeling.. How can i learn ngon topology what these means & how these works?? Can u suggest some channel for modeling??
If you're learning topology, definitely avoid ngons. I cover topology in more detail here: ua-cam.com/video/rwW6HpOcAuw/v-deo.htmlsi=7iXsTQeNBlLDnEg6
Also trying to get pockets and zipper flaps into pants is not only something I use to struggle with but still do. The problem isnt getting them to look right tho, its getting it properly rigged and weight painted.
i struggle soo hard with topology i can sculpt and texture no problem but man topology kills me sooo badly, It takes me out of the work creatively and grinds my production to a halt.
well i liked and subbed, this was great
The last tip.... the last tip just opened my third eye
Yeah, it's a game changer!
Hey bro make a new video on topology for hard surfaces and game asset it will be helpful 😃❤️
awesome tips
last tiip is groundbreaking I thought it was illegal
IT SOLVES SO MUCH PROBLEMS
Same here 😅
so just model the base form, smooth it then add more details, and then smooth again to finish?
question for add detail in subdivided level.
it's depend on the context right?, and for that case, it's for the model meant to be baked
if it's not the case, the lowest subdiv still need to capture the silhouette right?
hello and thank you for this great tutorial. i have a question : if i use quad topology and triangle topology in my 3D model, is this wrong? my 3D models are rims for marketplace and when I use the bevel modifier and select the arc option from the geometry section and miter outer section, the topology becomes n-gon and quad. that I have to triangulate them in order to correct them.
But the person who buys it can easily apply the Sub-D modifier on it so that the topology is completely quad and there will be no problem.
I am still having issues when it comes to the wavy pattern on cylinders when baking, most of the time 32 edges is enough to prevent that, but when i want to go lower like 24 or less edges, it just wont work. I have read the solutions to this online of baking with more edges and then removing edges after and that has worked in the past but for some details that more complex it just isn't feasible. If you made a video on baking cylinders, that would be a lifesaver for a self-learning guy like me wanting to model environment art for games.
You’re talking about the triangle fans on the end faces? The simple fix is to add one or two supporting loops close to the circumference. That will fix the wrinkly-subsurf problem.
I feel your pain and info through this regularly when baking cylindrical objects. I definitely plan to cover this in an upcoming video soon!
@@OnMars3D yeah, also the super black highlights or the opposite effect of white highlights when trying to bake a high to lowpoly soda can around the rim for example is something I’m trying to figure out. For example the new Spider-Man game has a lowpoly soda can and I wonder if they even used a high poly because their can is like 16 sides but has no baking artifacts from what I can tell.
@@OnMars3D thank you for replying!
Woaw amazing for all that, I'm a junior that just started in the vfx/animation industry and all those tips are gold mine for me, I see myself so much in your last tip ahah Thanks !
Congrats!! Hope your job is good!
@@Golan_Vivaldi Thank you ! I really love the few job I had so far !
3:36 I was looking at these images and what I find strange is that the details that are additive are not part of the base mesh, meanwhile the negative details are. But in the subd2 mesh, all the details appear to be part of the base mesh including the additive details. Almost like they are combined with a boolean then beveled. How was this done?
if you look more closely, the details on subd2 mesh are indeed separate, you can actually boolean them, but it would take more time and skills to pull off. As to why they look connected, I think it's just clever trick of the light, and maybe ambient occlusion. There is a trick that I normally use to fake a connection, which is to extrude out the edges so that it barely touches a surface underneath, if that make any sense
MY issue is I still haven't gotten the whole support loops and redirect edge flow down. still tryna undeerstand it and know when to add and when not to add
Can somebody explain the last tip?? U subd first - apply it, then add geometry and subd again?
tq
Personally I do retopology for optimisation, as my laptop and stuff can't bare having 100k tris on screen
Blender Tips: "Don't use edgeloops"
Literally every single youtube tutorial: "Just add more edgeloops"
How are we supposed to learn bro >.
can you talk about character game rig vs vfx rig
5:04 I'm very interested in this. Do you have video on this? Also, my only criticism for your video is, you going SUPER FAST over important things. Would've been nice if you slowed down and present the tip by simple/quick demonstration instead of listing a grocery list.
As a beginner I didn't understand anything what you are saying. So I guess i'm doomed to fail.