if i may butt in here a bit, the reason why jonathan and myself get so intricate with our models is to show that we can recreate a piece using reference. I suggested he used an odd sided cylinder because when he hides that one odd sided edge on the back of the granade, the granade's low poly silhouette ill look a lot smoother since is not symmetrical anymore.
Snader - for portofolio purposes, we try to get very low in our tri budgets. And yes, this may be argued farther along...but lets just let it die here! :P
and by the way, those were good things you pointed out!
Hey everyone! back with an update. I'm working on learning how to use this Marmoset, spent a couple of restless nights trying to learn how to do different things. I owe a lot of thanks to this amazing site! Still SO much to learn about normal mapping! http://wiki.polycount.net/Normal_Map
Here is my first go at Marmoset engine. I still have a lot to do with this bad boy. especially texturing. I did point out several things I know I still need to work on. So if you see anything else that I could use some more loving. please do point them out! Thanks guys!
I added the smiley face for fun... not apart of reference. Also will be making that pop out more when I give more detail to the normal map so it doesn't look so flat with the texture.
ah man... Im having lots of fun with this Marmoset now. This is the coolest thing ever! Here is another update. still haven't done several of the things I'm going to get to.. but I just had to see my progress and I'm loving it!
Alrighty then! I believe I am done! Learned a lot from this grenade! Looking forward to applying what I have learned towards the phone now. I have already done the low poly of the phone... will have a posting of that as soon as I lay out the UV's nicely.
Ok, so here is where I am at with testing what I got so far done on the baking process of the phone. Not added the alpha for the cord yet. I just wanted to see how everything was coming out so far. Pretty happy for the most part.
not bad. you've got so many waves u could sink a whole island with that! lol. Also, there a lot of part where u need to worry about the outside silhouttes 1st, and the worry about the inisde. TLike the cylindrical parts of the inside phone.
No offense, but if you've gotten better results with 10 sided geometry then 8 sided geo, you've screwed up somewhere. I went ahead and spent 30 minutes on these 2 models.
Left one has 1 bump per poly, gives really nice, even results.
Right one would/should give nicer results but that would involve tweaking the cage some more(you can see some shading errors on the lowest bumps).
I understand you want to have some inside modeling for "seeing the clip fly off and seeing that detail." But I'm gonna argue that you'd get a lot more visual effect from the silhouette unless you only have the insides on screen, and then your 500 poly model isn't gonna cut it either. You've got the pullring on a plane, but spent about 100-200 polies on the innards? That's triple the amount of your cylinder (4x10x2=80 triangles) , triple the amount of my left model(5x8x2=80 triangles), and quite a waste.
@raul I have nothing against low polycounts, I have something against polies spent in places where they're not that necessary. If you're gonna go for lowpoly like a flat hull, do it properly and decimate the whole model to about 200 polies. Because the difference between these models would hardly be noticable ingame, if at all:
BTW those statements contradict eachother don't you think? Making something very intricate and going as lowpoly as possible.
On the phone, you might want to read through this thread: http://boards.polycount.net/showthread.php?t=50516 and some similar ones. The central cylinder shape thing looks like the top in the first picture, and it'd look better if it were like the bottom of the first picture.
You might also like to add a few more polies to the lower rings. The top one has just as much sides, but has only 1/8th of the diameter or something. And you could optimize the bells with one of these sphere variations:
They might be useful for the 4 spheres at the top too.
I'm not sure if the circular extrusions on the top box are extruded, or separate elements. I'd go with elements like this:
Looks nicer, and saves a whopping 2 triangles per side. Plus makes it easier to bake.
Do the gears have to rotate? It might be worth it to combine the top half of the large gear with the housing.
Thanks Snader for your time of posting. You made a lot of great points. But I'm not exactly trying to go as low as possible here. The goal was to stay under 500 for the project and no more. Plus this is a portfolio piece that can look a little bit nicer and paying attention to details. I'll go back and fix the texture of the grenade later on when I have time to focus on my texturing skills. Now if the goal was to stay under 300... different story. If your given a limit amount, I personally suggest sticking with that and make whatever it is your making look the best as possible with that limit... Unless you absolutely don't need too... but thats up to you.
I know you aimed for 500. I did that with the first one too, but Raul said something about "try to get very low in our tri budgets" so I went and made that was very lowpoly. My point was, with both these posts, that the detail levels are inconsistent. You have a very bland cylindrical part, which is fine if you're going for lowpoly, ingame, optimized, whatever you want to call it. But the insides are very very detailed.
If you're going for a nice and detailed portfolio grenade, that's good, but this one isn't detailed. It misses geometry for some really large shapes. This isn't about not having detail, or having too much detail, it's about wrongly prioritizing detail. Instead of clinging to the polygon budget, aim for bang-for-the-buck. You stayed inside the 500 limit, but that forced you to sacrifice a lot of exterior detail.
Consider some of these things. You don't have to implement them, but think about it:
...does the firing pin actually need to be 10 sides, and have that... nipple?
10sided+nip = 60 tris, 8 sided cylinder = 28 tris, removing the backsides -> 22 tris -> 38 tris saved
_
..\ do you need the 9 loops on the striking lever, or could 3 loops show enough detail?
...| 9 loops 9x3x2 = 54 triangles, 3x3x2= 18 triangles -> 36 tris saved
Does the internal housing really need to be curved at the front, does it need those cuts at the back? Taking them away would save another 25 triangles or so.
That would make for a total of 38+36+25 = 97 freed up triangles. You can do a lot with 97 triangles, thats about 20% of the budget. For instance you could make pyramids of the bumps instead of flat planes, and then have some polies to spare.
Also, keep in mind that polycount budgets are ESTIMATES. Unless you're working with an engine like quake (whose *.mdl format which won't support more then 2048 triangles) or something else with fixed limits, polycounts are just a rough number. Of course you shouldn't make a 10k poly grenade, or a 10 poly grenade, but ingame performance is based on more then just polies.
Consider for a moment this: a wall that's 500 polies with a diffuse/normal/spec, versus a 320 poly sphere with parralax mapping and alpha transparency.
Also, here's a rule of thumb I use for my own work: Take the given poly budget, and aim for 10% below that. If you make it, cool, better performance. If you don't make it, you've got the 10% of breathing space. If that still isn't enough, you can go a bit over budget, assuming your other models do manage to stay in 90-100% of the budget.
johnathan-say thank you!! lol
snadder - that's a lot of great info. Thanx for taking the time to actually show some really interesting stuff that i wasn't certain about the low poly process. Those 2 links are pretty awesome as well. We arent exactly making these for games, and i cant tell u how many times people have bashed me for the grenade i made. Stuff like " oh ur tri count is too low, and oh your textures are too freaking big" , and blah blah blah. In the end, once i explain to them why i did it the way i did it. Then they give me a break. From what ive gathered, is good to have a low poly budget, and if not meet it then stay under it while still maintaining the low poly's silhouette as close as possible to the high poly. Correct?
So I am back with another post! started on this 3 days ago and am about 95% done... still have little things to do like finished the straps, the mounted receiver thats attached to the helmet, couple of wires... just little annoying things like that. I am tired and got bored and felt like posting it! Hope yall like! Let me know what ya think! Ill finish it tomorrow for sure...errr.. today... later on today! . :polytwitch:
Ged - Thanks man! I guess when I do the low poly I will create the huge textures at first and then make copies and make a smaller set.. depends on what I am told what to do for the project. It is a portfolio piece.. so yeah, Im sure i will make different layers and large textures.
One of the things i would suggest, is that you start putting a specular value on your material while you model. Also, there are a lot of edges that could use a bit more love ( bevels! )
Right on!. I think You should model as much as you can for the high poly so you can show it off. Then for the low poly, keep it to a low poly and optimize and cut everything together for a single final asset. one map. All that will still be a portfolio piece. This kicks the shit out of my helmet! lol great piece!
-Phil
Phillip - Thanks for the advice! I am still new to all this low poly stuff and what a single map is capable of. But when I do start on the low poly, I'm taking your advice on that. As for this kicking your helmets ass... I wouldn't say that. Yours is freaking awesome. Your detail in that eagle is amazing. and besides, two different helmets. Like comparing a scifi movie like ALIENS to an action movie like 300, both equally bad ass!
You do realize that Snader probably won't reply to your threads anymore because he gave you awesome critiques, that took him more than half an hour (probably way more, if it where me anyway ) and you didn't respond to it much... you first misunderstood him (he meant you could optimize a lot, not go lower in polycount) and then ignored the comment explaining that.
don't mean to get mixed up in shit like this, but that's pretty rude if you ask me.
Japhir - I have no comment on jasper, I have what I feel to be a good reason to ignore what he posted. Thanks you for the comment though! I just got back from class and the teacher pointed out a few things I need to fix
I sincerely doubt you could have any reason at all to ignore that advice, let alone a good one.
I sincerely doubt you could have any reason at all to ignore that advice, let alone a good one.
Yeah, he really talked about what you should be focusing on with a 500 tri budget,
This isn't about not having detail, or having too much detail, it's about wrongly prioritizing detail.
in the current state of video game engines, normal maps can give you a lot of surface detail, so the silloute becomes one of the most important parts about low poly models.
Looks nicer, and saves a whopping 2 triangles per side. Plus makes it easier to bake.
@Snader: Any particular reason you rotated the top cap slightly and triangulated the sides? I figured it was to round out the silhouette a bit but I made both models here for comparison and didn't really see any difference.
My guess is that it tricks the eye into seeing it smoother than it actually is because there are no shared parallel edges between the top and bottom. Similar in principle to using 5 sided cylinders as opposed to 6 sided for arms and legs on lowpoly characters, 5 sided appears less geometric to the eye than 6 and therefore the model looks more organic.
@Snader: Any particular reason you rotated the top cap slightly and triangulated the sides? I figured it was to round out the silhouette a bit but I made both models here for comparison and didn't really see any difference.
I feel it makes the model a bit more regular, because the triangles are all the same shape
/\ instead of some |\ and some \| To be honest though, it was mostly a gut feeling. I'll post a thread in Tech, with a more detailed analysis.
oops, I meant snader, sorry. no idea where that came from. And stop putting words into my mouth people, I didn't say it wasn't good advice. I said thank you to his posts, but for that specific little project on the grenade I told him that I had a 500 tri limit and I decided to go all the way with it to make it look nice. I personally do not see those posts that he put up shine on his own work, so it very difficult for me to really take advice from someone(from what it looks to be) that just reads other peoples forums, takes the knowledge of people that have experience and then to copy and paste that knowledge on other peoples threads. Sure he modeled a grenade really fast.. but still didn't prove to look better than mine, Cause hey, I said from the get go that I had a 500 tri limit. He ignored that statement. So yes, I have a very good reason to ignore his posts. Besides, the grenade was the first thing I have modeled in almost 4 years of being away from 3d. The dang thing took me 3 days to make the high poly cause I had a hard time remembering how to use Maya.
Can you not read at all?! Geez. For a 500 tri budget it does NOT look as nice as it could and he thoroughly proved you hat you could have improved the silhouette and possibly modeled the pin instead of using an alpha'd plane for it and still keep it within the budget. If you plan on improving at all you have to learn how to take criticism. Not every bit is valid 100% of the time, but this is seriously good advice that should not be ignored. You have used the budget very inconsistently and the best you could do would be to acknowledge this and take it with you to your next model if you absolutely don't wanna work on this anymore, but for fuck's sake don't keep the attitude.
Replies
Snader - for portofolio purposes, we try to get very low in our tri budgets. And yes, this may be argued farther along...but lets just let it die here! :P
and by the way, those were good things you pointed out!
Here is my first go at Marmoset engine. I still have a lot to do with this bad boy. especially texturing. I did point out several things I know I still need to work on. So if you see anything else that I could use some more loving. please do point them out! Thanks guys!
Left one has 1 bump per poly, gives really nice, even results.
Right one would/should give nicer results but that would involve tweaking the cage some more(you can see some shading errors on the lowest bumps).
I understand you want to have some inside modeling for "seeing the clip fly off and seeing that detail." But I'm gonna argue that you'd get a lot more visual effect from the silhouette unless you only have the insides on screen, and then your 500 poly model isn't gonna cut it either. You've got the pullring on a plane, but spent about 100-200 polies on the innards? That's triple the amount of your cylinder (4x10x2=80 triangles) , triple the amount of my left model(5x8x2=80 triangles), and quite a waste.
@raul I have nothing against low polycounts, I have something against polies spent in places where they're not that necessary. If you're gonna go for lowpoly like a flat hull, do it properly and decimate the whole model to about 200 polies. Because the difference between these models would hardly be noticable ingame, if at all:
BTW those statements contradict eachother don't you think? Making something very intricate and going as lowpoly as possible.
On the phone, you might want to read through this thread: http://boards.polycount.net/showthread.php?t=50516 and some similar ones. The central cylinder shape thing looks like the top in the first picture, and it'd look better if it were like the bottom of the first picture.
You might also like to add a few more polies to the lower rings. The top one has just as much sides, but has only 1/8th of the diameter or something. And you could optimize the bells with one of these sphere variations:
They might be useful for the 4 spheres at the top too.
I'm not sure if the circular extrusions on the top box are extruded, or separate elements. I'd go with elements like this:
Looks nicer, and saves a whopping 2 triangles per side. Plus makes it easier to bake.
Do the gears have to rotate? It might be worth it to combine the top half of the large gear with the housing.
If you're going for a nice and detailed portfolio grenade, that's good, but this one isn't detailed. It misses geometry for some really large shapes. This isn't about not having detail, or having too much detail, it's about wrongly prioritizing detail. Instead of clinging to the polygon budget, aim for bang-for-the-buck. You stayed inside the 500 limit, but that forced you to sacrifice a lot of exterior detail.
Consider some of these things. You don't have to implement them, but think about it:
...does the firing pin actually need to be 10 sides, and have that... nipple?
10sided+nip = 60 tris, 8 sided cylinder = 28 tris, removing the backsides -> 22 tris -> 38 tris saved
_
..\ do you need the 9 loops on the striking lever, or could 3 loops show enough detail?
...| 9 loops 9x3x2 = 54 triangles, 3x3x2= 18 triangles -> 36 tris saved
Does the internal housing really need to be curved at the front, does it need those cuts at the back? Taking them away would save another 25 triangles or so.
That would make for a total of 38+36+25 = 97 freed up triangles. You can do a lot with 97 triangles, thats about 20% of the budget. For instance you could make pyramids of the bumps instead of flat planes, and then have some polies to spare.
Also, keep in mind that polycount budgets are ESTIMATES. Unless you're working with an engine like quake (whose *.mdl format which won't support more then 2048 triangles) or something else with fixed limits, polycounts are just a rough number. Of course you shouldn't make a 10k poly grenade, or a 10 poly grenade, but ingame performance is based on more then just polies.
Consider for a moment this: a wall that's 500 polies with a diffuse/normal/spec, versus a 320 poly sphere with parralax mapping and alpha transparency.
Here's 2 good reads:
http://www.rsart.co.uk/2006/11/20/how-many-polygons-in-a-piece-of-string/
http://www.rsart.co.uk/2007/08/27/yes-but-how-many-polygons/
Also, here's a rule of thumb I use for my own work: Take the given poly budget, and aim for 10% below that. If you make it, cool, better performance. If you don't make it, you've got the 10% of breathing space. If that still isn't enough, you can go a bit over budget, assuming your other models do manage to stay in 90-100% of the budget.
snadder - that's a lot of great info. Thanx for taking the time to actually show some really interesting stuff that i wasn't certain about the low poly process. Those 2 links are pretty awesome as well. We arent exactly making these for games, and i cant tell u how many times people have bashed me for the grenade i made. Stuff like " oh ur tri count is too low, and oh your textures are too freaking big" , and blah blah blah. In the end, once i explain to them why i did it the way i did it. Then they give me a break. From what ive gathered, is good to have a low poly budget, and if not meet it then stay under it while still maintaining the low poly's silhouette as close as possible to the high poly. Correct?
-Phil
Here is my finished high poly helmet yall!
don't mean to get mixed up in shit like this, but that's pretty rude if you ask me.
The helmet looks very cool.
I kinda disagree on having the whole render 1 solid color. It helps.
Japhir - Thanks you for the comment! I just got back from class and the teacher pointed out a few things I need to fix
Titus S- thanks dude! Yeah, I personally like the green also, plus hey... THIS IS POLY COUNT!!! GREEN RULES!!
OOPS!! haha.. just realized I logged into my old account and posted this comment....
I sincerely doubt you could have any reason at all to ignore that advice, let alone a good one.
Yeah, he really talked about what you should be focusing on with a 500 tri budget,
in the current state of video game engines, normal maps can give you a lot of surface detail, so the silloute becomes one of the most important parts about low poly models.
@Snader: Any particular reason you rotated the top cap slightly and triangulated the sides? I figured it was to round out the silhouette a bit but I made both models here for comparison and didn't really see any difference.
I know, that was sort of rhetorical sarcasm, if such a thing exists.
I feel it makes the model a bit more regular, because the triangles are all the same shape
/\ instead of some |\ and some \| To be honest though, it was mostly a gut feeling. I'll post a thread in Tech, with a more detailed analysis.
Can you not read at all?! Geez. For a 500 tri budget it does NOT look as nice as it could and he thoroughly proved you hat you could have improved the silhouette and possibly modeled the pin instead of using an alpha'd plane for it and still keep it within the budget. If you plan on improving at all you have to learn how to take criticism. Not every bit is valid 100% of the time, but this is seriously good advice that should not be ignored. You have used the budget very inconsistently and the best you could do would be to acknowledge this and take it with you to your next model if you absolutely don't wanna work on this anymore, but for fuck's sake don't keep the attitude.