Project Post-mortem

A lot went on with the film room project and as a bonus we broke the sun. So this is something to sum up the project and how I think we managed.

Initially, like a lot of groups this project didn't kick off well for us. Even with lighting problems and data lose picking a room to work with was still the biggest headache by far, it took a long time to sort something suitable and suitable in this case mostly meant interesting and achievable.

Concepting wasn't really given the attention it deserved, since we were working with already established designs we didn't really allocate official time into our planning for the concept work despite knowing how important this step is, a lot more time and effort will be going into this from now on.

There were several improvements I could have made to my work I feel. The bar asset for a start even after remodelling it had a quite noticeable difference in size of the red part of the underside of it. This causes a noticeable difference in our image to the original once we set the camera angle.    While I’m happy with my work with the emissive material the colours of the signs needed some adjustment such as the red in the sign on the window was much too light and would have benefited from being darkened. Finally the outside scene I was not particularly pleased with, it was interesting to see what could be achieved with just normal maps, smoothing groups and metalness maps such as with the center building but the basic geometry I used is still quite evidently basic geometry. It’s texture could have also benefited from a little more time working with it to try and take away some of the look of blocky colours.

Overall I am happy with the project, the team worked well together and people worked hard. There where a few issues with our finished scene that didn't match up with our image

-Emissive lights in the dividers and walls were too yellow

-The outside buildings looked like they were too far away

-The perspective of the camera was not completely correct

-The shots didn't line up exactly

We had hoped to address these issues but a data loss and lighting problem meant we had to prioritise recreating the scene again.

While our teamwork was overall good in my eyes it did seem Thomas ended up doing a lot more work with modelling and engine as all the assets had to be added into his building scene. It would have been better for the rest of us to have been able to pick up some of his work load.

"Sharn, we've got some bad news...."

I made a rookie mistake with my last blog post, I said the project was going well and we were on schedule and nearly finished which naturally opened the flood gates for disaster.

On Monday afternoon having just finished explaining to a tutor how well everything seemed to be going the rest of the team appeared with excellent comedic timing to tell me that it had broken.

Nobody knows what happened but the short version of the problem is we lost the sun, no idea where it went and it didn't have any intention of coming back. Having left Unreal 4 idle and to it's own devices for a few moments it seemed to have misplaced the skybox which was a fairly big problem to begin with but this also broke the lighting we had set up in scene and turned the metal textures black again. In the end the only option we had was to rebuild the scene in a new file. On this day we learnt a valuable lesson about backing up files.

On the plus side when we did rebuild it the bar fit in a lot easier with the new remodelled version, the with the improved UV unwrap the textures didn't blur and stretch on the model and this also left me with room to dedicate more space to the part of the counter that would have the shared bump map with other metallic objects so that stopped looking completely awful. Also while looking at the original scene idea I noticed that the far end of the bar was capped off by an extra piece of the counter that was actually fairly visible so I modelled and added into the asset as well. 

Looking at the version I had and the remodelled version it's quite clear what the problems we were having with this model were caused by, those original corners are pretty poorly made. The bend modifier had been giving me a lot of trouble so at the time achieving those curves had been a victory but returning to this modifier for the remodel I had a bit of a eureka moment and realised a simple solution had been staring me in the face, so I moved the pivot point a few centimetres and it was fixed.





As you can image this week had suddenly become very busy so when the sun left he took my earlier plans to work with blueprints with him but while that had to be taken off the table I was once again troubleshooting in the material editor when connecting up sign texture in scene provided a new hurdle. Since the material was a shared texture sheet and a lot of information in it was controlled by the Albedo map such as the emissive colour we had a few problems with glowing windows which were just a little more than noticeable. So I set about getting a mask working within the material. I'm not actually too sure about how technically accurate it is as far as materials go but it did the job I wanted so at least it's not wrong.

So in the end, the end being 9PM the night before hand in, the scene was up and running again but we couldn't quite get the lighting and camera angle back to the level we were happy with like on the original file so it was never going to match up perfectly with the film still but it came out alright all the same and in terms of group work the project was a success in my eyes, everyone did their part and more and we produced a finished scene, twice technically. I've really enjoyed this project more so that I thought I would, the team aspect made working a lot more fun and kept me motivated, I'm sad to hand it in and move onto the solo sentry gun project.

So this is the final scene we ended up with. There are lots of improvements that could be made, as I mentioned the lighting and our camera angle isn't quite right. My textures on the street outside also look a lot plainer and block colour than I'd like there are also a couple of errors on the texture that slipped by our notice until we were staring at it on the presentation screen. Again, not perfect but we did alright.

Go team!

It’s been a busy week for the group, 10 – 9 stretches in the labs, problems with perspective and asset sizing and an unholy list of bad puns in an attempt to settle on a team name but by Friday night we had something to show for it, a scene in engine with the main bulk of its assets in place.

We still have a few things to do, some more assets need completing, some changes need to be made to existing ones including the bar I made which needs a remodel on the curve to fit the scene. I’ve also noticed a few problems with it I want to address while I’m there and finally of course, the lighting needs building.

Having gotten a fair bit of our stuff in engine on the Friday we’re working a little bit ahead of schedule which has given use a small window of time to go back and improve the accuracy of the models.

I started working in engine a little earlier than everyone else to get the neon signs working but in having achieved that I seem to have been dubbed ‘the closest to a lighting expert we have’, this isn't even close to true considering the signs use emissive textures and not lights but eager to remain helpful this weekend has been a crash course in lighting and using Blueprints in Unreal 4.

Searching the internet didn't turn up a lot of information or tutorials for what I was looking for so through trial and error and a rehashed UDK set up I managed to create a material that flickers for the sign in the window. The original set up had a second branch connected to it which causes more variation in the flicker but the result seems a bit too fast and excessive so it was simplified again.

I also sent this week working on my third task for the scene which was the outside view from the window. Originally we'd intended to have this part be just two painted planes but with room in the budget we decided that creating some simple geometry would be preferable so be able to capture shadows and a sense of depth that wouldn't be achievable with a flat image.   

It was interesting to see what could be achieved with just smoothing groups and normal maps but there where still issues with this approach such as our view from the original image window was blocked by characters so there was a lot of estimating. The box geometry was useful for this too, with the ability to move the position of the buildings we were able to make adjustments to try and force the perspective to match the image.

So far our group work has been going well, everyone has been working well together and we're pleased with the results so far and we're hoping to be finished very soon. 

Week 2: Dex's Diner

This week was all movie surfing and mood boards, endless mood boards. With the Asset swap completed everyone moved full time to the next which was the "Film Room".

"Exhibit A"

The target for this project is to recreate a film still of a room from a movie, matching perspective, assets, style and importantly the lighting. The largest hurdle we've faced so far is picking a room to work with, after the groups were enlightened about brown and green colour palettes last week there's seems to have been a mass attempt to go out and find the most colourful films imaginable seemingly to the point of madness with some. By the end of the week the tutors goals of getting us to think more about our palettes was pretty much realised and at some stages, obsessively so.


Eventually, this was the scene we landed on.

Dex's Diner, from Star Wars attack of the clones. There were a lot of factors to take into account that led this to being our selection but we thought it would be a good scene to work with even with the problems we might face with the perspective.

Starting with work load, all the scenes we collected either provided too little work for our group of five, or too much. This room is full of repeating assets meaning only one of them would need making. With the scene divided up among the group each person is only required to produce one large asset and two small ones apart from the person making the building structure, since it's a fairly large and complex job it will be their only modelling task. My own tasks consists of creating the bar counter, the neon signs and the outdoor view.

Since a large part of this project is looking at lighting we also thought this scene would be useful to work with as it is lit by such a large natural light but also has the influence of artificial ones which would force us to work harder with and gain a better understanding of Unreal 4 if we wanted to get it working.

Not ignoring the fact it's also cool, because diners are cool. They were cool in Hopper's Nighthawks and they're cooler when you put them in space. If we're recreating something then lets recreate something interesting.

Overall this week as we get into our modelling work I'm surprised to find how much I'm enjoying group projects, everyone learns quickly from each other and because of that each member of our group is coming away from these projects with a handful of useful tricks and skills they've picked up from someone. For someone like me who likes to keep to myself it is a very strange sensation to be eager to take part in the group. I was expecting and dreading the second term big group project but these smaller ones have been a huge help. Thanks course restructure! 

Let's go image hunting!

Taking a lead from our new optional class of “Cool Shit Wednesdays” I thought I’d take an opportunity to try and interact with my blog more and I’d make note of some of the interesting art I’ve been hording away in my reference folders this week.

While a large percentage has been mostly Russian Mass Effect art I’ve also picked up some more constructive pieces while browsing the Polycount forums and in particular some pieces I found on a ‘women in game art’ thread (http://www.polycount.com/forum/showthread.php?t=140562)  

[1]

I'm starting with this one, mainly because I like looking at the topology of other peoples models. Especially with characters the topology of my work isn't always particularly neat and I find having references of what I should be aiming for helpful for improving. 

[2]

This picture I just really liked, something about the art style and colour palette just felt very art deco even if it did take a few moments to notice the very large sword embedded in the unfortunate fellow.

Looking into this I found out it was art based on a game called Transistor from the makers of Bastion which I intend to get my hands on if I can find ten spare minutes.

[3]

Last image I think I'll throw up here will be this one because I enjoy isometric art styles, since they're fun to look at and fun to draw. I always want model toy versions to pick up and build towns with. 

This one I like in particular because of the overall shape and the little decorative touch of the pieces of cog on the walls.

This seems to be a style that keeps popping up in mobile gaming so I guess it's an art style we're likely to see more of.

Well I'm going to leave this post at that, just something small to break up the week but I'll probably add a few more of these posts when I go off on another image hunt and find more things of interest.

[1] http://firstkeeper.deviantart.com/

[2]http://jenzee.deviantart.com/

[3]http://catell-ruz.tumblr.com/page/2

Week 1: Game Art Round 2

I’m back at DMU for second year and another round of game art and with painful memories of a stress filled, sleep deprived final hand-in week still fresh in mind it’s pretty safe to say I have a much better work ethic and frame of mind this year.

But with welcome backs short lived we were thrown in the deep end not only with two group projects but the discovery that while we took a break the games industry did not and the diffuse and specular maps have been usurped in favour of PBR, physically based rendering. So here’s your maps, there’s Unreal engine 4 and this is the asset swap, enjoy!

The asset swap being our first project so to give everyone a taste of team work with each team member completing a step in creating an asset then passing it on. So I spent my first day spending a disproportional amount of time painting pictures of benches before a warm reunion with 3DS Max, which promptly crashed.

Once it was up and running again however I was working on a somewhat odd task of modelling a roasted pig, which I must admit is not how most vegetarians plan on spending their days but never the less I think he turned out ok and fairly cute given his circumstances.

 I chose not to name him knowing he would go on to be unwrapped and cooked by my teammates so I instead passed him on and moved onto my next tasks of unwrapping and texturing.

 I've done very few purely hand painted textures and the wood of the shield was generated with Photoshop filters but in this case I'm saying it counts but regardless I was very pleased with the outcome though my texture for the copper could have used some improvement, especially once it was in engine and my attempts to come to grips with the new roughness map left it particularly shiny and reflective but other than that it went alright.

As for the second group project there was only one topic of conversation:

Brown. Brown and the realisation I have never watched a movie that wasn't brown.
This project was to recreate a film still in engine but avoiding the recurring colour palettes of green and brown, a harder task than you might think. So we're still looking for something to work with on that front.

More on the visual design side life drawing is very rusty in terms of form and drawing speed so that's going to require some work to fix it.

All in all it's good to be back and I'm looking forward to this year, here's hoping I can organise myself better this time around.