Friday, 15 May 2015

Alice in Wonderland; Post Mortem

It's important to reflect on projects, to think through and document all the lessons learned- both practically and emotionally, and now that we've had a week to cool off from our hand in, it's time to do just that. Our Alice in Wonderland project was the largest group project we've been allocated; spanning 10 weeks (+3 unofficial Easter weeks) and it was also the first time we've been dictated random groups in accordance to our specializations. I think this is where I learnt the most- not necessarily in technical skills where I feel I've made bigger leaps in other projects, but rather in management and communication. Especially so, since this is the first group I've been in where it's been an issue. I'll revisit this shortly though, let's start with the positives.

What went well is the overall atmosphere we managed to create in our two, highly contrasting levels. I think our forest beautifully managed to strike a balance between a grounded Oxford forest, and still have that Alice in Wonderland flair- through the use of assets like playing cards, books, and clocks as platform pieces intermingled with the vibrant greenery of the large leaves in comparison to our shrunken Alice. We accomplished some nice vista shots, that I like to believe are largely down to my highly detailed composition planning of the forest level. Especially so since this was an area that I conceptually worked on alone and was responsible for the design of the most of the assets within it.

 It was a strange system though, where our original three levels were divided up into three people managing them respectively. Myself for the forest, Jake for the Hallway, and Christy for the Canal. The dynamic was effective and efficient during white boxing and early engine/mechanical stages and worked nicely as a system, where I felt in control and the workload was divided up equally between people who were conformable with engine.  However our system fell apart heavily when it was deemed necessary to take the engine files and merge them into one cohesive level. Logically this makes sense and I accept that it was required, but once it was done my input on the area I had brought this far was taken out of my hands and my input became more verbal, aside from the occasional day where I would be able to work with it personally. 

Aside from that this had a huge knock on of side effects, it meant that Jake was now managing all three areas which is a huge amount of responsibility, and there wasn't much we could do to help, since only one person could access it at once, but this meant that the fine details in our levels fell apart a bit since it was too much for one person. For example I made 4 rocks that could easily be manipulated in size and scale to form nice walls and rocky banks almost seamlessly, but as the engine person wasn't as familiar with how they worked as I was, I feel they were laid out in a way that didn't maximize their potential. I mentioned it a few times, but it was a low priority task and by the time it became and issue, they were all already placed and it would become a huge job to reorganize them. I know other people felt the same way about their assets as well, but because it was all funneling through one person, time wasn't allocated to nicely placed population as much as it probably should have been. 

It's easy to say that we could have avoided this by keep the engine files unmerged, but this would likely also lead to more problems later down the road,  instead trying to merge already delicate and nice engine files together. Not sure how- but I'm certain something would have gone horribly wrong here. What we could have done is just have days where the engine was free and accessible to everyone to just adjust, fix and make small changes to their assets and the level before it became too late to change. 

The fact that I only now realize people had a problem with the way our group was structured speaks largely for itself, the group wasn't hostile or uncooperative, it was simple generally splintered. Problems that people had weren't addressed and because of this communication suffered and the overall quality of the level went down. The ironic part is we were all in labs everyday, and mostly the same room- and this was still an issue. My problem wasn't specifically with communication since I had a lot of contact with the team, both teaching and learning. I should mention that everyone was great at being prepared to listen and adapt to new techniques, as well as sharing freely any information or tricks. I think Christie stood out as someone who taught the group a lot of technical skills. No, my problem was with formal criticism. I think that while most of us talked a lot, it generally wasn't as formal as it could and should have been. It felt more like everyone was just 'updated' and aware of what people were doing, but not criticizing or being critiqued. In future there definitely should be a daily meeting where we talk about how to improve everyone's current assets. 

What I think worked well was the quality standard that people reached for and attained. While our art style changed dramatically from the ‘set’ style, everyone adapted as we went along and we still arrived at the same style, but without all the stylistic trims we’d planned to do. This was mainly due to how the ideas changed while we were whiteboxing. Despite the change, no ones assets were out of place or looked like they didn't belong, this was mainly because we were abiding by a realistic style standard. However once again, while the assets people made were realistic, sometimes- and I'm guilty of this as well- took criminally long. I'm going to specifically mention Sarah and Hannahs Rabbit and Caterpillar. I don't think these assets were relevant enough in the level to take up the time spent on them, which was the majority of their time, even though each character was only on screen for about 12 seconds each! The problem though is that they wanted characters for their portfolio (as they are charter artists- which we had three of) on this was their final project of the year- which is entirely fair. Our decision to let the caterpillar and the rabbit become major modeling characters was a mix of obligation to abide by the specialization list, and the knowledge that in these early stages of the project we wanted our group to fundamentally work happily together. It seems unavoidable really, you can't expect someone to not want to spend eleven weeks working on something that's going to help them get an job/internship in the department they want. 

I feel the best way to have worked with this would have been to set much stricter deadlines, because while there was deadlines on texturing and certain assets, they were all entirely ignored and this was never addressed as a group. People just worked at their own pace which left us in the frantic mess that we were at the end of the project. We definitely needed stricter deadlines.  

We had to cut out level down in raw length multiple times this project, using only a third of the original cave plan, and cutting out the canal entirely (which was a sad day for all) and our level was still too long. We should have started with a much smaller concise level plan, and just expanded after the DMU hand in, for the separate Off The Map hand in, which we still have a long time for. This would have meant we had a high quality level to showcase our artistic abilities, then afterwards, we could work more on having a fancy game. The fact is, by the time we knew we had a level that was too big, all the individual stages had already been worked into too much for it to be worth cutting. The work we would lose just wouldn't compensate for the extra man power dedicated to one level with the remaining time frame. All we needed was a better schedule and less characters being made. 

Integrating mechanics naturally and organically into our level was also something we had relative success in. The falling leaves and house of cards were two simple puzzles I made that helped the player feel involved with the level without breaking the flow of the game- it also meant I had a blast stacking physics based cards for a few days. The more complicated puzzles in the hallway that Jake and Christie did also hit their mark as they were all very integrated into the scene and felt like organic puzzles. 

In fact as an entire game it felt very fluid and cohesive, aided by dynamic cameras- although maybe too many for my liking- and matinees for cut scenes and explanatory purposes that made the level feel more like a game.  I especially like the underwater cut scene and the teapot one, as they were both integral for explaining the scene. They also looked cool. Props to Jake on that one. 

On a personal level, I'm pleased with the quality of the assets I created, and also the processes I learnt- there's more detail on that in prior blog posts. But despite this, and the fact that one way or another- everyone worked really hard on this project, I just don't think it lived up to our expectations. It's hard to point out whats makes the level so disappointing for me, individually everything looks great but I think the we just missed out on the fine details, and ironing out the strange camera and buggy behavior.  It's a shame because I was proud of the project until about the last week, when it struck me just how much work still needed to be done with it. It was a challenge to stay positive after that. I also think every was just tired with it in the last few weeks, peoples motivation was lower, especially after our character redo and essay hand in.  I think the most important thing to learn from this project is organisation and scheduling, in groups I've been in we've managed to get away with only having a loose schedule because we worked so well as a team and were constantly helping and being involved with each other. It wasn't the case this time, so I feel we needed stricter deadlines and to be formally critiqued on our assets on a daily basis. 

My action plan for Summer is pretty solid, I'm already stuck a new student contest that I'll talk in more detail about another time, and I've got a small game art team together working on making a fully functional co-op platforming boss fighter game- we even snagged a programmer for that one. So yeah, summer will no doubt keep me busy and focused. In terms of Alice we're unsure. The plan was always to work past the deadline but now that it's passed and the general negativity about it, I feel everyone, or at least most people, will be to burned out to continue. We've decided to give it a few weeks to take a break, and see how we feel after that. 

Sunday, 12 April 2015

Once More, with Colour

So I'm revisited the rocks, after a hearty break and mushroomy fun. I couldn't help but brood over what woes these rocks would bestow upon me this time around.
It actually went totally fine and was a relatively painless process. 

Basically since I'd worked out all the hard stuff in my last confrontation with them, most of the issues were already ironed out. On top of that I had my poly painting experience with the mushrooms to fall back on, and overall it was a relatively painless experience. The only thingy worthy of note, is how lucky I am that it did go so well- because unlike with the mushrooms, this time I was completely reliant on the poly paint to work, since the rocks had been unwrapped in zbrush, so while they were perfectly technically even, trying to hand paint the texture sheet would have been impossible to work out. My last resort would have been using a tillable texture, since I was aware of the possibility coming in, but luckily it didn't come to that. I much preferred using a polypainting method for these assets as the rocks have lot's of small details and dents that I was worried wouldn't quite come through in the normal map, and wanted to pronounce these features through the diffuse map as well as the height.
Overall I think they came out pretty good. There's two concerns I want to address overall about the rocks though. One is that because I didn't end up using the original assets I modeled for the rocks, there was no need to be constrained to their shape. Making rocks from scratch in Zbrush is easy and simple and I think I could have gotten way more out of the shape if I had gone down that route from the beginning. But because I sculpted these rocks with the original intentions of baking them down onto the low poly, I never explored the shape as much as I could have. If I have to do rocks again, my pipeline would go more along the lines of: 
Model a high poly in Zbrush > Decimate a low Poly version > completely retopologise that version in 3DS Max or 3D Coat > Unwrap that in 3DS Max > Bake it down. 
I think knowing this process alone may have been worth all the trouble it's caused me, I've worked out a zbrush pipeline that works for me. 
The second thing I would address is putting more colour variation in the texture. The reference image I was using was just tone greys, but regardless I think I should have taken some colour liberties like with the mushroom and added hints of browns and reds to make them more interesting. 

I also worked on water, which was an interesting experience. I had to make 5 new bump maps to create the different textures, then base colours that change at different depths to create the refraction on shallow areas and darkness at deep areas. I thought the way it reflected light was really successful- in my unreal scene at least. Once again the problem of working on different unreal scenes is while my water looked great in mine, when I recreated the effect in the official scene, it just completely failed. It was dark and ugly. A problem that I could have addressed in the process if I had been able to work directly with the scene lighting. But all the tampering in the scene couldn't get my water to react properly to the conditions being used. It's out of my hands now, which is a shame, because I'd love to see it working. 

Sunday, 29 March 2015

The Mushroom Factor

Fun week, good stuff.
The natural progressive step for me was polypainting, I love sculpting, it feels organic and natural and fun, so it didn't surprise me when I found polypaining and equally awesome way to work.
I am still limited to British mushroom colour- red and brown gets old fast when you have 5 different mushrooms to do, but tried to find way of stretchy colour and pronouncing different hues, even if it mean exaggerating life slightly. I felt it was necessary since all these mushrooms would be so large in comparison to Alice, that the colour would need to hold up in interest and quality. Also, it's worth noting that like the rocks, I want the mushrooms to be as adaptable as possible so all the different heads and stalks are separate exports, and certain heads and stalks are made to be able to interconnect with each other. This is to help keep variance where we need to and make the most out of the assets we've made. Especially since the colour palette was so restricted, most of the mushrooms look somewhat natural when combined.

This is one of the nicest and most detailed texture sheets I've made. Each part of it has had a lot of care put into it, which is why it makes me immensely sad that I've packed such detailed assets so tightly. I think these mushrooms could easily have done with being two separate texture sheets. When these were all baked out I was sad to see the quality drop from the sculpt to the bake, purely through how the resolution of the texture. Negotiations are taking place currently as to not have the texture size reduced further, I'll fight for my mushrooms. 
In the meantime everyone's assets are slowly coming together, alot of the plant life Christy made for the Canal level are being put in the forest and it looks really cool. In the meantime however, the forest engine file I've been working on is being absorbed into the bigger game. I want to say it'ss too soon, but I think that's just because I've been enjoying working with my own file, although it will be considerably harder to work with it now that I also have to abide my someones schedule as well. On the bright side our level is getting sewn together, and we'll soon be able to play from start to finish. Yay!

Sunday, 22 March 2015

Rocky Trials

Trying to find a way to make these rocks work is a test in my patience as a human being. I've spent days now trying to use the sculpt without the original mesh, and it's been a huge trial and error process.
This is my first time trying to use a sculpt in zbrush that isn't just a bake, the hardest part is finding something that's  unwrapped. I've figured out decimating my sculpt, zremeshing, and unwrapping in zbrush to make an acceptable- but terribly optimized low poly rock to use as the base mesh. The problem now is unwrap space, I'm trying to import the obj in 3ds max to package it with everything else then use that fbx to bake the high poly onto, but for whatever reason that I cannot even begin to comprehend, xnormals won't align the high poly sculpt with the new unwrap. 

This rock is a time vampire and spending anymore time on it that I already have is going to cripple me emotionally. The only way I've managed to get this to work is using a low poly, decimated version with its own automated unwrap sheet all to itself. It's inefficient in both triangles and texture space, but I'm thinking I can just try and build extra assets into the texture sheet afterwards.

EDIT: OK! Some progress- I've found t hat once I bake it all out with 1 asset per texture sheet, I can then take that and then repackage it in 3ds max, under the strict condition that I cant reunwrap or change the unwrap sheet. This is still profoundly irritating but at least I can get at least to rocks on a single texture sheet. It's not much, but right now it feels like a good compromise. 

These rocks still need texturing, but I feel like the hard part is done. Did some mushrooms to decompress- making absolutely sure I liked the 3ds max models before I normal mapped them, in fear of having to repeat the same grueling process again. In contrast though, the mushrooms were actually therapeutic. 

Sunday, 15 March 2015

Alice in Wonderland: Rock Edition

Rocks are the foundation of our level, in both the cave and the forest areas. As the forest area is my primary concern, when I started these rocks it was specifically for that area- hence why most are flat, as they are meant to be by a river, so I was trying to implicate the smoothing effect the water has had on the rocks to an extent. My intention was to do another set of rounder, rougher rocks for the cave, however the learning process took longer than expected so unless I get more time later, these rocks will be used for both. I think there is enough variance in shape to support that at least. 

I started in 3ds Max, unsure of our polycount aim or budget, so I started relatively low. On reflection, the original models were never going to be enough to supplicate the quantity and size these rocks needed to be, as well as how often they would appear in the level to not justify a much higher poly count. Regardless I didn't realize my error until far later. 

 I modeled my rocks so they had a different shape on each side of the rock, so when it decorating the level, you effectively had 4 times as many rocks, also with the intention that you could stretch the shape and rescale things for even more variance, When I started sculpting some detail for a quick bake, I got quickly carried away with the detail. It was a internal conflict to make the rocks as beautiful and detailed as possible without breaking the original shape of the model. Otherwise it wouldn't bake. What I got was an ugly hybrid. I was too carried away with the sculpting that I broke the original shape too much that the normal maps didn't work. It also occurred to me how much potential they had, and how satisfying and easy it was to model rocks in zbrush. I instantly felt that my new sculpted rocks were wasted on their original models.
I've been looking into re-topology software/retpoloising in zbrush so that maybe I could recreate my rocks based on the sculpt, and rebake based on that. 

Sunday, 8 March 2015

Stacking Cards

So I spent this week stacking cards. Well not entirely, but it actually did take out a laughably big portion of my day- I think it was my favorite day so far. 

No but on a serious note this has been a solid week for me, I produced a fully functioning detailed blockout of the forest area, which included collapsible cards and a leaf falling puzzle. I also did quick blueprints for a double jump and bouncing mushrooms. They were both easier than I would have suspected- or maybe I'm just getting better at engine. Huh.  

Regardless, I made a huge amount of white box assets to basically map out the entire scene and all the unique assets we'll need to make it. This feels like a big jump in progression of the forest area and it means that forest assets can now be divided out and distributed. The forest has different areas of light and dark, you walk into a well lit open areas with sun rays lighting the character that contrasy from th incredibly dark and claustrophobic atmosphere of the cave. But as you travel thorugh the level, you also venture through shady and imposing sections- like the gnarled tree roots that block out the sun, as you climb up the tree and back into the light. As well as this there is verticality as you scale different areas then fall back down again to keep the path interesting and the player thinking, as well as areas of backtracking- from the teapot back through the log to the tree, which is now scaleable. 

I made a temporary stand in test of the leaf puzzle that basically functions as it should, to make sure it was possible to maintain a path that varies as different speeds of leaves reset at different heights. This meant doing a lot of individual matinees and lots of testing to make sure they always worked. Unfortunately since matinees aren't copy and paste-able, when we get a final model for the leaf it's likely that I'll need to redo them again. Which is fine, since they way it's currently structured isn't particularly tidy- it basically just works, but there's no finesse or leaf like motion to it. 

Sunday, 1 March 2015

Alice vs the Redesign

This week we talked a lot about the length of the game and where we wanted to focus our attention, at this point it became clear that Myself, Jake and Christie were in charge of the Forest, Hallway and Canal respectively. Based on this any questions on direction on each of the areas tended to be directed towards us. I think this made the group more stable as there was now some form of internal organisation starting to appear. As me and Jake are the most familiar with engine, this also helps since we would be the ones working most closely with the mechanical aspects of the level. I've also noticed that a general pattern in all the groups I've worked with so far, is that the engine person tends to be the leader of the group, just because of the funneling process and the responsibility of the actual 'construction' of the level. It's also up to you to know what is possible and what isn't and design around our engine skills.

This week we also talked about our long term goals, obviously we all want to win the competition, but it's also important that at the end of the 11 weeks, we all have nice portfolio pieces. The problem is that is our three character artists then all wanted to do characters- naturally. It's a strain on the group since we felt that the only character necessary was Alice herself, and the others could be done on stylistic planes. However it's unfair on our character artists if this is what they wanted for their portfolio on our final project of the year. Our decision to let the caterpillar and the rabbit become major modeling characters was a mix of obligation to abide by the specialization list, and the knowledge that in these early stages of the project we wanted our group to fundamentally work happily together. It does make me slightly nervous though, since it means only the minority of the group will be actively working on the environment.

We also talked this week about what major puzzles we wanted, for definite these were a rabbit chase, a size puzzle and platforming puzzles. The details aren't quite ironed out but just knowing what gameplay elements we're working with makes it easier to concept around.

The personal kicker for me came at the end of the week when Jake returned from his trip to talk with the people behind the Off The Map project in Oxford. He basically told us that they were looking for something strongly grounded in Oxford and the lore of the book. This didn't fare nicely for my magical, emmisive mushroom forest. Yep.

I needed to re-coat my forest in a fresh paint of traditional British realism. Right down to the mushrooms. 

RIP emmisve mushroom lights. Week 1- Week 4