Wednesday 29 October 2014

Film Project; Post Mortem

So with the conclusion of the film project, what have we learned? I think I've picked up more engine and technical skills in the last two weeks that I have in the entire first year. It was intense, thrilling and enlightening. I feel more confident after this project about my future on the course than I have done for a long time, and that optimism will hopefully be long lived- I just feel excited for the next time I'll get to sculpt in z-brush or make emissive lightning in Unreal 4.

What went well overall was our ability to work as a team. We were like-minded, ambitious and organised, using lists and graphics to keep scheduled- which wasn't much of a problem anyway since everybody was in labs everyday, often working from 10am to 9pm. This is also made communication fluid and constant, there as a constant stream of feedback and help where it was needed and noone was reserved about teaching processes or technical skills. It was a great environment and I'm terribly thankful to my team for such an atmosphere.

In terms on the scene itself I think the technical aspects were particularly strong. We were the only group to have an interactive element (the flickering, glowing sign) and that made it really stand out. We also had lots of texturing detail using things that are generally unstanderadized within our year group yet, like projecting mapping and z-brush that made our scene look polished and accurate.

I also think our worked well because of the huge amount of instanced and repeated assets, which meant we could make out scene bigger and more complicated than others, without crippling our tri budget.





I also feel our scene choice was quite strong, we were ambitious with it and I know some people regret not choosing a simpler scene and doing it to a higher quality. But on a personal level I'm happy we chose this difficult one because it had so many unique opportunities to make the most of metallic textures and a little bit of sci-fi architecture that makes it so interesting.

Aside from our great engine crash from a prior post, the  biggest problem we ran into was t he perspective and shape of the room. Dex's Diner is a rounded scene so establishing that was allot harder than we had anticipated. The bar in particular was a nightmare to fit, because it was so critical in the shot, it's position was paramount but because of the grooved shape, any mistake was glaring so it had to be constantly re-sized and shaped to fit the room. Lighting was also a challenge since we were dealing with such a huge amount of light sources and try to replicate not only the mood but individual shadows. However one always had an effect on the other making it awkward to balance. There was also so many different light sources to deal with- natural light from the windows, artificial lights and the other numerous different lights in the room and so on.

If we had the opportunity to start over, one largely agreed upon notion was that we should be more careful distributing jobs; Tom, who organized our engine (because he happened to organize the main room so assets were naturally sent to him) was inadvertently given him the biggest job. To avoid this in future I would perhaps have a dedicated engine person who handles less assets and texturing because of the extra responsibly they take on. We also could have had 'official' asset reviews. As I mention earlier we were constantly reviewing and feed-backing each other's work as we went along all the process- but an official review might be a good idea to really critic each individual asset as a group all at once, so multiple changes could be made simultaneously instead of a steady trickle of suggestions.

Overall though I'm proud of the scene we made. Working as a group was an incredible source of knowledge and the I think we all emerged better game artists because of it. Also on another personal note- this project has gotten me into great working habits- similarly to the last group project. My pace and workflow has improved, and the quantity and quality of my work has grown because of it.



Praise the sun(box)!

So...
Our sky is gone. Literally. We were all in labs working on our project making a very important group decisions- our team name (it's CTRL_ALT_DEFEAT by the way.) In a strange blip of white light, our computer screen that was hosting the scene, suddenly blipped white, then  black then...
                                                   Oh God, it's all... it's all broken 

Our scene collapsed in on itself like a devastating black hole, and destroyed all our lighting, textures and simply removed everything we had outside the scene, putting us in a void. We had no idea why this was happened (although we suspect it had something to do with skybox?) Regardless it meant we had to re-import every single thing again since autosave blocked us from reverting. It also butchered our metals and left them a dark grey, regardless of metal or roughness maps. 
There was nothing we could do to save or recover the scene. No matter how much the scene was lit up our metal was black.
The only course of action was to rebuild our scene from scratch- and to our dismay, as we began to attach materials in our rebuilt scene it happened again. Luckily the solution this time was to bring in scene reflection sphere to put light back into our scene. If the crash taught us anything it’s to not rely on autosaves and save many alternate versions along the way.  Also to never delete the skybox. Maybe. Who knows? 

Frankly I'm just relieved our team is ahead of schedule enough for us to rebuild in time for the deadline (in two days!) Though it is a shame, we were in such a comfortable spot and now its chaos and panic. Though at least this can be an opportunity for everyone to fix names and organize all our assets and textures, right!? and while that is true, this still sucks, massively. Bugs happen. 

ZBrush; my favourite worst enemy

So we've divided up our assets and I was allocated the stool, misc silver and cups because I was the only member of our group familiar enough with zbrush to tackle the cup designs, which we knew early, would definitely need to be sculpted.
 
I started with the basic cups at 300 triangles each (I used 252) to make each of the basic shapes. One was slightly higher and had a larger rim but aside from that they were very similar, but the slight difference is shape made a nice difference considering how many time the cups would be instanced and repeated around the room.

My next task was to sculpt the detail into the cups, I had a very limited experience with zbrush prior to this so I was slightly apprehensive but once I started I found myself enjoying the more authentic ‘modeling’ feel to it. It was far easier to creature the curves and grooves this way than in a normal map. Unfortunately while the zbrush object itself was overall successful, I had a huge amount of trouble high poly baking it because of some things I failed to prepare and anticipate I needed in the 3ds Max file.

We decided we were going to conserve texture space and combine the cups, ketchup, salt and pepper, table, scones and plate were not slated to share the same space. Unfortunately because I’d already unwrapped and sculpted based on that unwrap, it was now unusable. There were also some problem with the bake itself as the cup was not centered to the world when it was exported as on object (I thought I could move the bake once it was in the scene- but it doesn’t’ work that way) so I needed make sure for cup was in the exact correct final position before I took I into zbrush.  I also had a problem with dividing the geometry in zbrush completely obliterated my corners and made the edges and rims smooth- so when I went to mask out the top and bottom, I didn’t realize I was missing one of the edges because it had been totally smoothed out. The mask was also a bit rough as you can see because some of the pattern goes over where it shouldn’t.  So for these reasons the first iteration of the cup didn’t work.  Too help with the dividing I tried turbosmoothing the mesh before exporting, this helped with the divide itself and everything kept it’s angles, however it made the high poly bake process very tricky, I also couldn’t quite get the pattern the same and zbrush consistently crashed so eventually I decided to just export it as a normal map and just copy it into height map. This gave me the exact pattern I wanted from the original try and also allowed me to completely and cleanly keep my angles by editing where the lines ended and began on photoshop. The overall effect is incredibly similar to what a high poly bake would have achieved but was far better to handle.
  I tried to do a similar process with the second cup but this time I realized that because unlike the prior, this pattern was repeating and need to be even and equal, that it would be more appropriate to do it with the standard height map. This yielded a much crisper result and I now know where the use of zbrush is more appropriate.

Texturing the cup was frustrating, because the actual reference has absolutely no roughness or texture detail, I constantly tried to fight the urge to over complicate the design, often I added to many roughness details and had to backtrack because it simply didn’t match the image. Even if the end result looked good. A few iterations in I found a good balance in detail and just let my normal map be the prime decoration of the cup- just like in the shot. Although a higher resolution picture would have been a great aide. 

Everything is brown!

So our new project has been set! The task at hand is a group project to replicate a scene shot from a movie as accurately as possible. Sounds interesting.

Choosing a scene is the most important creative decision to make in this project- we need a scene complex enough to be interesting but not one that would absorb all our triangles or time. To compensate for this we wanted a scene that had incredible presence and mood, with interesting light and composition. With this in mind we made individual random mood boards that we then brought back together for discussion. We discussed how we felt about anime or cartoon rooms as some of them had interesting light and shape, but eventually decided we wanted to air on the side of realism. We made this decision because we didn't want to put something that had already been stylized and put in engine back into engine.  We also felt the scene that left more overall impact while still being feasible were the realistic ones.
Once we narrowed our choices the one that stood out was a shot form ‘Guardians of the Galaxy’ that I found, we felt it had an interesting challenge with the grids and the light shafts, as well as interesting texture potential and a shot that leads you to a solid focal point of the scene. The beams frame the light in the middle and we felt this would make for an aesthetically pleasing scene.  With this in mind we presented our idea to the 'dragons den' and were advised to reconsider based on the bland brown colour palate. We felt the texture and lighting were enough to make the scene stand out however soon resigned back to our search that combined what we liked from Guardians and a scene with an interesting colour palette.  
It was an important point that was made though, when we reflected on our original moodboards, they truly were overwhelming brown and dull. They had interesting compositions, sure- but did they stand out? Not really, not in the way we needed our scene to stand out. We were advised to look at influential artistic movies from the past like Space Odyssey, that was simple yet effective because of its contrasting colours. Also to disregard any personal attachment we had to the moves themselves, and look at them purely objectively and expand our search to movies we hadn't heard of/seen, or even considered watching before. 
Once we began searching again it was a far more arduous process and we looked specifically for bright and colourful palettes. The problem we had is while these scenes hit the criteria of an interesting shot, we didn't really like them, and often the colours were abrasive and unpleasant so we continued to cycle through.
We originally got to the point where we were choosing between two scenes: 
Both of these scenes were solid choices- once was hindered by over complication and the other by simplicity- we had just come to the decision to do the one on the left, when I stumbled upon our Star Wars scene, it had the style and character of a retro bar but with the sci-fi chrome textures and lights that I love so much. It was oozing with potential and there was enough individual assets to slate the needs of six ambitious group members. We returned to the 'dragons den' with our scene, armed with colour picks and grey scale studies. We felt it was the strongest scene we’d found that had exactly what we needed.  The colours and lights kept the scene bright and interesting enough without dominating it.  Another beauty of our scene is it uses so many repeated and instanced models which means our scene has the potential to look far larger and complex than our other choices. We knew the scene would be challenging but were more than prepared to tackle it regardless, the ‘difficulties’ simply served as an excellent opportunity to learn new technical engine skills.

To our enormous relief- the choice was approved and with apprehensive optimism, we move onto the next step. 

Sunday 12 October 2014

Team Not Applicable!


It's Game art year two and we've kicked off with some group work, it’s pretty crazy since all our lab groups have been sporadically moved and you've ended up in a group of six or seven people you don’t know  particularly well. At first I was apprehensive but it only took one day working with my new team to fill me with optimism for the rest of the year. After a short week my new group are also my new friends and I’m genuinely excited to continue working with these people on future projects- staring immediately with the Film Set brief.

The strangest part is I've found myself spending my days working in labs instead of at home, and that’s a huge, huge difference because I've learnt so much just idly watching whoever’s next to me and picking up on tricks and tips that have totally changed my workflow. It’s also far easier to just ask someone around you for help or for crits rather than to search fruitlessly online for a solution.  Also strangers who happen to see what I’m doing and compliment it or ask me how I did it is really helping my confidence and I’m finding it’s finally an environment I feel comfortable working in.

Another advantage to group work? It keeps me working at a pace that I wouldn't usually be able to maintain. Wanting to impress and cooperate with my team keep me on a strict schedule that in the long run reduces my stress and helps my general organisation and time management skills.  I’m working far more hours than I did last year and it feels productive and good. Like I ’m seeing real and significant improvement.
Two weeks into the second year and I’m loving it.


Anyway, the actual project itself was a Viking Tavern Bar scene which was pretty interesting for a quick thing, I designed and modeled this candle playing with the new roughness and metallic texture maps- which are delightfully effective. I think they really helped my candle model looking as real as possible. Oh and also the new engine- for all the complaining going around I think I prefer it.... or just don't hate it yet anyway So far it's eaten less of my work and hasn't made me murderously angry or chronically depressed yet so what's not to love? Though I do miss the prettier default sky-boxes from UDK.