Good to see you back on it and back into 3D! Looks like a good start, though I imagine it will take some cleaning!
Yeah no doubt, I did have a generous hand when building some of the parts of the station, but others I kept the poly count down. From this point forward I will be as cheap with the polys as possible, but I won't sacrifice sharpness for low polys. I actually thought of doing a video tute on Optimization on DS9 . . . what do you think?
gp
I think it's a great idea to share your thoughts and insights in optimisation, however, in terms of being instructional you need to bear 2 things in mind:
It needs to be generally instructional: I.e. it needs to explain why you are making the choices you are making so that people can apply that to their models, and not just to your very specific agenda for this model; I despise very specific tutorials which only explain how to do something, not why you are making the choises to do so, so that you can apply it to your own models.
and most importantly you need to think on how to make it not incredibly dull LOL: more than 50% of my modeling time is probably spent optimising the mesh after I have built it, removing redundant polys, merging some together, removing loops and in some cases splitting polys I have booleaned so they don't break later (never leave polys with 'holes' in them!) However while certainly theraputic and 'easy' to do, if mind numbingly slow and boring too, it might not make very interesting watching; it's not unlike watching someone playing a first person shooter over their shoulder, it can make you feel travel sick and you don't get the feeling of immersion they get haha You do need to think on how you can make your videos short and to the point enough so people can learn without getting bored of watching it
Neither of those things are easy, but both will mean it'sof the most use for others! In either case though, making tutorials is always a great idea to help people, and even the seasoned hand can learn new things!
Yep . . it's nice to get the juices flowing again . . .
I started in on some of the detail work, already this piece of the mesh is up to 9.500 poly's . . and this is as low in destail as I can get it and still keep it looking cool . . . as much as I try, if I want the detail in, the high poly count is unavoidable
started back on DS9 again . . I finished the modeling portion of it, and I also went back over the entire model and did as much optimization as possible. as of now the entire mesh stands at 5 million Poly's.
At the moment I am in the texturing phase . . . uv maps and procedurals are going to be the method of choice.
here are the updates . . .
gp
can't help but notice you said 5 million polys after optimisation, I know it's high detailed and all but 5 million seems really excessive, like tehbian's pluto station is 7 million and it's covered all over in unique greebles and has a full interior, perhaps that's the problem, your greebles can be cloned around and cloning often leads to massive poly counts
I see greeble pits dotted around the model, it's hard to tell without a wireframe but I'm guessing that's where most of the polys have gone, but they are small areas in comparison with the rest of the station, so I think their poly counts should be proportionate
Thanks for the welcome back . . .
IRML, yes this is very true, there is over 1.5 mil in greeble poly's on the docking ring alone. Those sensors are expensive, but well worth it.
On the other hand, I've gone through and pulled out all of the backside polys, a lot of the waisted tube polys that were in the docs and made a lot of the smooth rounded piping jagged because of the reductions . .
I think over all I eliminated over 700,000 poly's across the board. What's left is large numbers of polys in detail bevels. Example: The square boxes around the windows on the habitat ring all have multiple bevels Those boxes alone are 228,990 polys, each box containing 273 polys. Now I could just have a plain old square box around the window and cut the number from 273 to 5, but if you did a close up render of the window, it would look like 5 polys. That is what the entire station looks like nice detail.
I think with today's machines this number will not be an issue when it comes to rendering. Ive rendered the entire mesh on my machine in 32 bit mode with BG radiosity and it took a few minuets . . . besides the model will be constructed in a way that whatever part of the station isn't being used in your shot, you don't have to load it which will make life easier . . .
nodoubt there is always some more randome polys and if I rebuild a section here and there in a different way, I can cut down the count even more . . but after 4 years of construction I aint rebuilding $hite, it goes as is
gp
Great model! How are you finding handeling a 5 million poly model with a 32 bit version of Lightwave? Are you using most of your ram or do you have quite a bit of memory left?
it's true high poly counts don't do as much damage to render times as people think, well in lightwave at least
I get what you mean about not redoing things, but some advice about about the bevels for future models, a single bevel will work fine most of the time, getting you a look closer to a high res bevel than no bevel, but a poly count closer to no bevel than a high res one
IRML, I totally see what you're saying definately, I will keep that in mind, after having to optimize this model brought about a new appreciation for the finner points in beveled edges
I see the examples you sent of beveled edges, to be truthful I didn't bevel anything like that for edge smoothing my bevels were mostly design issues, which I am going to refrain from in the future "trust me" . . here is a shot of the windows so you can see what I spent my poly's on
gp
ah ok that makes more sense, when you said it could have been a 5 poly box I thought you meant each window was like a plain box, so I thought you spent that 270 polys just on the bevels haha, but you have made much better use of them
Yeah but still, is all that extra design really necessary? sure there could be less bevels around the circular window curve and do I really need all the other levels . . probably not . . . ! live and learn I guess.
but a close up of the window shot will be interesting, next model I will be more frugal.
gp
texturing has come into focus, . .
a fellow forum member, Rigel, suggested I go a little less dark with my texturing . . I think I will take his advice.
gp
Thanks dude . . yep, the colors are close enough, Only Paramount can tell you for sure what the colors are, everything else is speculation.
gp
I think the red is too saturated. Try to tone it down a little, and break it up by giving all those greebles in there different colors (different shades of that red, and also sometimes gray, brown, yellow).
I don't think the greebles are textured yet
Awesome work, great to see you back at this
Correct, haven't gotten to them yet, at the moment just trying to get through the pylons, The Sensors (greebles) are a special case. I have to texture those in two sections, the base (greeble pit) and the greebles themselves. the reason for that is that doing a far render of most of the station one may wish to leave the actual greebles out as to conserve on the poly count, but the greeble pit needs to look as if the greebles are in it. That will be the case with all of the sensor pits, just to give people the choice either leave the polys in or take them out.
thanks for the tips tho . . I will keep them in mind when I start mapping them.
gp
Posts
Yeah no doubt, I did have a generous hand when building some of the parts of the station, but others I kept the poly count down. From this point forward I will be as cheap with the polys as possible, but I won't sacrifice sharpness for low polys. I actually thought of doing a video tute on Optimization on DS9 . . . what do you think?
gp
It needs to be generally instructional: I.e. it needs to explain why you are making the choices you are making so that people can apply that to their models, and not just to your very specific agenda for this model; I despise very specific tutorials which only explain how to do something, not why you are making the choises to do so, so that you can apply it to your own models.
and most importantly you need to think on how to make it not incredibly dull LOL: more than 50% of my modeling time is probably spent optimising the mesh after I have built it, removing redundant polys, merging some together, removing loops and in some cases splitting polys I have booleaned so they don't break later (never leave polys with 'holes' in them!) However while certainly theraputic and 'easy' to do, if mind numbingly slow and boring too, it might not make very interesting watching; it's not unlike watching someone playing a first person shooter over their shoulder, it can make you feel travel sick and you don't get the feeling of immersion they get haha
Neither of those things are easy, but both will mean it'sof the most use for others! In either case though, making tutorials is always a great idea to help people, and even the seasoned hand can learn new things!
here is a quickie update . . . Verti-bevel is our friend
gp
Good to see this project and you back.
Yep . . it's nice to get the juices flowing again . . .
I started in on some of the detail work, already this piece of the mesh is up to 9.500 poly's . . and this is as low in destail as I can get it and still keep it looking cool . . . as much as I try, if I want the detail in, the high poly count is unavoidable
gp
At the moment I am in the texturing phase . . . uv maps and procedurals are going to be the method of choice.
here are the updates . . .
gp
can't help but notice you said 5 million polys after optimisation, I know it's high detailed and all but 5 million seems really excessive, like tehbian's pluto station is 7 million and it's covered all over in unique greebles and has a full interior, perhaps that's the problem, your greebles can be cloned around and cloning often leads to massive poly counts
I see greeble pits dotted around the model, it's hard to tell without a wireframe but I'm guessing that's where most of the polys have gone, but they are small areas in comparison with the rest of the station, so I think their poly counts should be proportionate
IRML, yes this is very true, there is over 1.5 mil in greeble poly's on the docking ring alone. Those sensors are expensive, but well worth it.
On the other hand, I've gone through and pulled out all of the backside polys, a lot of the waisted tube polys that were in the docs and made a lot of the smooth rounded piping jagged because of the reductions . .
I think over all I eliminated over 700,000 poly's across the board. What's left is large numbers of polys in detail bevels. Example: The square boxes around the windows on the habitat ring all have multiple bevels Those boxes alone are 228,990 polys, each box containing 273 polys. Now I could just have a plain old square box around the window and cut the number from 273 to 5, but if you did a close up render of the window, it would look like 5 polys. That is what the entire station looks like nice detail.
I think with today's machines this number will not be an issue when it comes to rendering. Ive rendered the entire mesh on my machine in 32 bit mode with BG radiosity and it took a few minuets . . . besides the model will be constructed in a way that whatever part of the station isn't being used in your shot, you don't have to load it which will make life easier . . .
nodoubt there is always some more randome polys and if I rebuild a section here and there in a different way, I can cut down the count even more . . but after 4 years of construction I aint rebuilding $hite, it goes as is
gp
I get what you mean about not redoing things, but some advice about about the bevels for future models, a single bevel will work fine most of the time, getting you a look closer to a high res bevel than no bevel, but a poly count closer to no bevel than a high res one
I see the examples you sent of beveled edges, to be truthful I didn't bevel anything like that for edge smoothing my bevels were mostly design issues, which I am going to refrain from in the future "trust me" . . here is a shot of the windows so you can see what I spent my poly's on
gp
but a close up of the window shot will be interesting, next model I will be more frugal.
gp
a fellow forum member, Rigel, suggested I go a little less dark with my texturing . . I think I will take his advice.
gp
gp
I think the red is too saturated. Try to tone it down a little, and break it up by giving all those greebles in there different colors (different shades of that red, and also sometimes gray, brown, yellow).
Awesome work, great to see you back at this
Shame we will never see a new DS9 on the big screen or the small.
Correct, haven't gotten to them yet, at the moment just trying to get through the pylons, The Sensors (greebles) are a special case. I have to texture those in two sections, the base (greeble pit) and the greebles themselves. the reason for that is that doing a far render of most of the station one may wish to leave the actual greebles out as to conserve on the poly count, but the greeble pit needs to look as if the greebles are in it. That will be the case with all of the sensor pits, just to give people the choice either leave the polys in or take them out.
thanks for the tips tho . . I will keep them in mind when I start mapping them.
gp