Discouraged on Making High Poly - Looking for Modeling Workflow Advice
polycounter lvl 8
So I'm not necessarily a beginner, though still a lot to learn, but I'm still getting discouraged while working on some projects. I thought I might see if there is any advice on workflow, or general advice, when it comes to the high poly model of a high to low poly process. I've done a lot of modeling but never took it so seriously until recently and it's seemed to kill a lot of my flow. The focus on workflow might be getting in the way of actually creating something that looks good.
So I was wondering how do you guys handle the workflow of high poly modeling? How do you decide how much detail to add in the model and at post when texturing? How do you handle such a large amount of polys? Also what about the fact that a model might look bad during the process? The model looking bad during steps has sometimes caused me to go back and make changes even if it would look good once the step was done.
I used to just create models without thinking too much about these details and by my current skill level they turned out great. Now that the questions are there creating models has become much more stressful and I don't have any answers. Any advice is extremely appreciated!
So I was wondering how do you guys handle the workflow of high poly modeling? How do you decide how much detail to add in the model and at post when texturing? How do you handle such a large amount of polys? Also what about the fact that a model might look bad during the process? The model looking bad during steps has sometimes caused me to go back and make changes even if it would look good once the step was done.
I used to just create models without thinking too much about these details and by my current skill level they turned out great. Now that the questions are there creating models has become much more stressful and I don't have any answers. Any advice is extremely appreciated!
Replies
I could always make characters for rigging that deformed well, and spent a lot of time learning how to rig cause I am a megalomaniac and need to do the entire thing from sketch to running around in an environment. Pretty early on I got excited about displacement maps because they made sense, so I was already doing half game models: high poly in ZB and low poly cage, extract displacement maps, but the whole game character pipeline was a mystery. I couldnt figure out how to bake AO and normal maps so I did a workshop with Jon Rush which removed all doubt and put me on the right track.
Once you have the basics right it is just a question of augmenting your workflow with new information that can be found here on polycount. If you are having trouble with the process I would recommend taking part in such a workshop, it will straighten out all the kinks.
Cheerio
I've been slowing learning zmodeller and it certainly is a nifty tool, the fact that you can stay in one app is super important. But i also think it's really important to lean how to model in your standard 3D app. Personally it helps me with blocking out, keeping things clean and most importantly its accurate. Zbrush as far as I know doesn't even have a standard scale. Transpose to flatten out surfaces accurately? no thanks.
With that said blocking out meshes in zbrush for retopo is great.
Ok so here are a few things that I found to help me when it came to modeling in Maya. (im sure the principles carry over)
First one. Be aware of how you are moving your points. Is your axis world, object, component, etc? Why struggle with a certain way to move a vertex when you can change the way to move it.
standard way of moving it, as you can see its not precise. And for me causes a lot of frustration.
changing the way you move it. (ah man sorry you need to open image in new tab for this one)
And some other things.
All in all, its just a lot of practice and finding out why you are getting frustrated at a certain process and trying to make it less frustrating. The examples i gave may not resonate with you. That's fine. Figure out what works for you.
"Also what about the fact that a model might look bad during the process? The model looking bad during steps has sometimes caused me to go back and make changes even if it would look good once the step was done."
oh man, haha I suffer from this a lot. I guess i'm an instant gratification kind of guy and my patience needs work. Thing is if you are baking a cake. And you eat it before you finish baking it...well...Its not going to taste good. So that's the same principle haha. If you are modeling something and you haven't even finished it. Hows it going to look good?
This is a part of why I created this post. I know a bit more than the basics, I know how to model, but now because I know what I'm doing I see all the things I was "blind" to before, e.g. what I'm doing "wrong". Improving my workflow so there is not a constant restarting would be great. I haven't been able to find that information on here as you mention. Anything you would suggest looking at?
Arkaria said: That's basically what I'm going through too; periods of time without making anything. I'm not sure how to go about creating without mental road blocks. That's why I'm hoping to gain some suggestions and ideas for different workflows to try. Although like you say making something that isn't my best is probably better than nothing at all. Thanks.
Wow, that's a really great example; eating cake before it's done. Instant gratification is nice but often when you let something bake as long as it needs to it turns out best. I need to remember that metaphor.
Also the example with the storm trooper helmet is good. The idea of blocking out hard surfaces and then going back and cleaning up the topology makes a lot of sense. I'm usually trying to make the high poly with good topology from the beginning and it causes me a lot of headache. That also kind of ties with what Arkaria said about just making something that isn't perfect. Then in this case going back and cleaning it up. Also using edge crease until you finish and add the support loops makes sense. Those are some really smart ideas I'm gonna have to try. Thanks so much for those examples.
And this is what learning and practice is all about. It's one thing to be prepared and have ideas solutions before diving in, and another to have the experience to predict what solutions you'll need. I like to spend time now and again reinforcing little hard surface shapes; just doodle out something interesting but also trying to challenge myself by throwing in some divet or panel line in a way that doesn't line up with my geo, for example.
As far as "looks bad in the rough", it's important to keep your blockouts, reference and end goal in mind while thinking a few steps ahead.
Another tip for creasing in Maya is that you generally want to work low, and use PageUp to bump up the subdivision level a notch or two, it's much comfier to work this way. You may need to bump back down when it gets more complex and start replacing with support loops, but that's down to choice/cpu/ram/style, I think.
The techniques they use can often be applied to what you're doing or they at least insert ideas into how certain forms can be achieved in an effective and manner that doesn't induce mental strain. My favorite person to watch is Tor Frick, he was the technical director on the latest wolfenstein and does some crazy awesome work. He has a youtube channel where he posts his streams and they're super insightful and enjoyable to watch. https://www.youtube.com/user/Askguden
I couldn't find a good way to bake textures and I wasn't sure about constraints. The workshop I took dealt with the mechanics but we got a lot of information about making effective game characters as well. I see you are busy with hard surface and I reckon by the sounds of things you would be up for a Gnomon tutorial on producing the kinds of assets that will make you happy. I notice with those DVD tutes that they dont only stick to the mechanics, there is also lots of info on aesthetics too.
There are so many ways to do it, if high-poly modeling for the purpose of baking: edge/box modeling sub-Ds, dynamesh-zremesh, Zmodeler, hard surface sculpt-slice curve-zremesh, hard surface sculpt-very simple manual retop-crease-subdivide, meshfusion-retop, open subdiv, spline-based polymodeling....even good old booleans and round edge shader method.
There are just so many methods/techniques that it's great to know them all and just get in there and model using whatever gets the job done.
Also, something else that I think could have an impact on modeling methods is the increasing popularity of VR content. Normal maps just don't translate too well in a VR environment, coupled with ever-incresing vert count thresholds could pave the way for a major shift to straight polymodeling, a workflow that would essentially render the need to create a full high-poly model obsolete.... Who knows?