Second opinions about this hard surface topology?

Alex_J

Hi guys, I got some feedback from some professionals about some work I completed recently. One thing that was pointed out specifically was this weapon. They said that the topology was a no-go as it was not clean and wouldn't suffice in production environment. I am gonna ask them directly for any further clarification, but there's a lot of great hard-surface gurus here so I want to get some varied opinions as well.

First, let me explain how I made this weapon.

I began by getting a reference. In this case, I had a nice photo that was at a flat angle so I just put it on a plane in maya, made the plane live, and used quad draw to trace it directly. I adjust proportions a little afterwards just to make it feel better. That initial blockout was built in all quads, designed to be cleanly subdivided. Support edges and all that -- your basic subD modeling. The blade recieved two levels of smooth, and the handle I took into zbrush to add engraving and some IMM ornamentation.

Then I took that initial blockout and turned it into game model. I stitched off all edges not holding silhouette defining shape. So the blade is flat on either side except a mild taper on the inside, which is why I need the one edge going down the middle. Everything else can be reduced to only essentials.

I did the bake, came out good. You can see the final textured result below.

In your opinion, where can my topology be improved? Would you say it's a no-go? I'd appreciate it a ton if you can explain why. Some examples why this wouldn't work in a game, or might be a pain in the ass for other parts of the pipeline. One thing I always try and do is have clean edge loops running the length of an object so that you can split into UV shells very easily. Not always possible but I did so for this model. Blade splits into two sides and the handle as well.

UV's. I kept the handle at the same texel density as the blade, but this is a mistake because the blade does not have enough detail on it to show a discernible difference even if there is some texel density disparity. So I could have sized the handle shells up more. Anyway, just wanted to show that I design the topology with cleanliness in mind.

 

Snweos

There is generally a couple of things you want to keep in mind when it comes to modelling:

1. Does the mesh need to deform at any point?

2. If you are using a subdivision workflow, are there nice edge loops and topology for it to divide correctly?

3. Will the mesh cause any issues with vertex normal shading?

4. If the model is being used for a game, is the triangle and vertex count too high?

5. Is the model not a complete nightmare to unwrap?

---

In relation to this particular model I would say:

1. No it probably doesn't need to deform.

2. Apparently you did factor in having nice topology for the subdivision workflow, though it's not one of the screenshots shown.

3. It's hard to tell how clean the normal shading is with the texture, but it looks fine for the most part.

4. I would say the poly-count is kind of fine, certainly a few areas I would definitely optimise, but I tend to work with mobile / lower end pc, so everything tends to look pretty high too me.

5. You already unwrapped and textured the model, so I assume you were fine on this front.

---

At the end of the day as long as the final result is displaying fine in-game or in a render, and it's not too much of a pain to make changes to for yourself or other people then the topology doesn't really matter a great deal.

Noors

No-go seems really arsh to me. What explanations did you get ?

Topology wise, i'd say you have a lot of definition on the handle and not as much on the blade curves. So this should be more consistent. Too much poly for the handle ? Depends of the specs... view model, world model ?

The blade looks dull and doesnt really tell a story. if i type kora sword, i see the blade is actually thicker. It could have been more interesting if the center "fold" was more marked. Not sure what ref you had. But that's not purely topology related anyway.

What are the tetris shapes ? Long objects could be mapped on a 2/1 texture ratio aswell.

musashidan

Are you sure these so-called professionals knew that this is a real-time game asset topology they were looking at?

Yes, you could certainly do some further optimisation, but from a strictly topology perspective(I agree with Noors on Uvs/blade design) just blankly writing off the topology as unacceptable, without knowing its purpose or destination, just seems elitist to me. 

Alex_J

Appreciate the responses guys.

The reviews are from a mix of professionals who are not necessarily games industry pro's, but also cinema and all sorts of disciplines, not just game characters. But I think they make a lot of reviews and have to filter tons of student art, so totally understandable that they see something that at first glance looks like decimation master soup they aren't gonna sit there for five minutes and really dissect it. There is a lot of other stuff besides this particular weapon in the scene.

But I just wanted to make sure I wasn't being either flippant nor that my method for making this wasn't completely idiotic.

@Noors , the tetris blocks are the balls on either end of the hand guard. They are just smoothed cubes, so I left the default unwrap as it worked fine in this case and made decent use of space compared to spheres. About the design of the blade, yeah I totally agree with all of that. I wasn't super happy with this piece but I had too much on my plate and just needed to finish at a certain point.

sacboi

Second Noors & Musa.

I also scanned your char WIP a month or so back for a sticky beak, really come up nice man! cheers to that. Can see the end result for all the effort put in which IMHO the sword can definitly benefit from as a purely aesthetic point of view. When or if time permits of course.

Alex_J

sacboi said:

I also scanned your char WIP a month or so back for a sticky beak,

THanks man.

I had to google sticky beak. Lol, learn somethign new every day. I wish 'Merican had fun terms like this.

Noors

well yeah sure if they were expecting a subD model for a movie...

poopipe

OK, I'm going to play the bastard...  Apologies in advance.. 

OK. Brutal bit first.. 
If this was the most recent piece in your portfolio I would not invite you for an interview.
 
I don't mean any offence, I feel it is extremely unhelpful to be indirect. 

The distribution of detail is very uneven - the blade is the biggest bit, why is it so much more faceted than the handle? You could get a better silhouette by moving geometry from the handle to the blade and incur no overall cost. 
How well does the blade shade? I suspect badly based on the mesh and the shaded screenshot. Evenly spaced strips would almost certainly shade better, any extra cost incurred could be saved elsewhere. 

UV questions... 
Do you really need unique UVs for each side of the blade.. Can you see both sides at once? 
Why aren't the  shells for the handle straight? 
Why didn't you unwrap the handle as a straight cylinder? 
Those dirty great T shaped shells are way too big and  why aren't they overlapped? 

Honestly the UVs just look lazy to me and that in itself is enough to put me off a candidate 

I won't go into the material work because that's my specialist subject and I don't feel like typing that much. I would back up everything Noors said on that though. 

Advice: 
Work out what your view distance is and prioritise the distribution of mesh detail according to that. 
Do the same with your UVs and for the love of all that is holy re-use texture space where you can. 

Again, I'm not shitting on you for fun. those are the sort of initial assements I make on a lot of pieces that come past me.  

I do agree with everyone in that the feedback you got was pap - they were probably just busy.. 

Alex_J

Thanks @poopipe , you don't got to apologize i appreciate you taking the time.

Your assessment is right. I was lazy with this prop. (basically I was lazy with the technical details of most of the props in the larger scene).

Truth was this is one of the last props after spending upwards of nine months on the rest of the scene, so I did just lazily unwrap it and that was that. Initial plan was for many of the props to be neatly packed in an atlas together, but that kind of went to shit because I was figuring out a ton of stuff as I went along. Major thing that detracted from quality of this piece is that I was figuring out the design as I went. That kind of makes it impossible to know exactly how you will organize things, and eventually I just ran out of steam and started treating the props like prototypes and not really finishing them out as proper game assets. I am aware of and understand all the concepts you mention, I simply abandoned the extra work of proper optimization with a lot of the props -- there was six (though I made several others that ended up getting scrapped) and I made them after two full characters so there is very much a failure of endurance (or poor planning, rather).

I simply took on too high quantity so eventually I had to cut down on quality to get done.

The questions about why I didn't straighten the shells and why not mirror some parts -- that was because it has unique engravings on both sides. The detail is not symmetrical. I did not straighten them because I didn't want my tiny little engravings to get distorted at the the edges. Yes, I realize it is a poor judgement overall as the handle is hidden while the blade is prominent. Not just that, but there was no need to for such high amount of unique and tiny details. It mostly goes unseen.

I hadn't shown breakdowns specifically because I was not very professional with the materials and UV layout. THe character and tiger are pretty clean and follow standard practices but beyond that I was lazy. I simply bit off more than I could chew and got to a point where I just needed to finish, and since I was giving so much time and effort to all the unique detailing I cut corners with UV's and overall optimization.

So if you were a recruiter and asking me, "are you lazy?" The answer is no, of course I am not lazy, I just didn't have enough experience at the time I made this project to realize how much work it would end up being. I attempted a higher quality than anything I had done so far and also much higher complexity and scope as well. Too much challenge. Eventually had to cut corners just to finish.

Lesson learned is scope better. Basic beginner mistakes I'm sure you've seen many times.

P.S. if you got some time to type out your views on the materials sometime I'd really appreciate that.