Week 11: Character Duo Post-Mortem

Just before we break up for Christmas our dichotomous character duos are finished and handed in. Just like with the turret project, this blog post will cover the entire process, what worked  and what didn't work.

Concepting

The entire project started with me just thinking of existing duos, and then making a pinterest board out of it. While I was doing this I was also using my notebook to think of ideas of my own duo; this was a mistake but I didn't realize it at the time. By picking a theme and style this slowly began to pigeon-hole my thought process which eventually led me to a potentially stale design.

I had eventually made a choice on the theme and began to design the first character, which was my second mistake. I should have designed both characters together. I struggled with the initial shape generation of this character which is why I jumped straight into sketching heads.

With one character done, I began to think about the second character. I knew I wanted this character to be a larger, protector of the child, yet not combat focused. I was confused on how to go about generating ideas for this character, so I tried what worked with the turret, a bash kit. I grabbed shapes that I liked from the moodboard to get my bash kit going.

It was at this point that I had to present my concepts. I had only one character complete with a small selection of silhouettes for the other character. The end result of the presentation was that the silhouettes are far more interesting than the final character, and while I liked the child soldier, was the idea really pushing myself out of my comfort zone? I decided to scrap the child and restart with the silhouettes.

Post-Presentation #1

I pulled more shapes from the moodboard and just began piecing them together to make interesting shapes. I didn't think about the duo or the theme at this point, I just wanted interesting silhouettes and eventually I had a good selection to chose from; then I paired them up into interesting duos to try and get some good dichotomy going. The silhouette I picked had elements that reminded me of feudal Japan, with a headdress and lanterns so I began researching the theme on pinterest while trying to push the idea within the silhouette.

I picked the duo I liked the most, which was an older, larger guardian figure and a younger girl, who clearly travel together. With a couple picked I pushed forward with the design process until I had orthographics.

Modelling

As we only had to model one character I chose the big guy as he seemed a lot more interesting with his huge backpack. I decided to go for a 8,000 tri-limit, which was larger than we're used to but not completely insane. It felt like a comfortable size. I modeled it with Z-Brush in mind, learning from my mistakes on the turret but this also tripped me up, as I found out later, I didn't focus on the overall model enough, my mind was to focused on the sculpting.

Base model complete I took it into Z-Brush and began sculpting. This time I actually had fun with Z-Brush, as I wasn't learning it for the first time. There's still plenty to learn but I got stuff done! I then baked it all out, added the extra arm and the other lanterns and then dropped it into UE4. I still needed to the pose character but I was getting ready to begin texturing.

I had to do another presentation for critique at this stage, and unfortunately things didn't quite go to plan, which is something you don't want to happen with only a week left. The overall critique was that my model didn't capture the likeness of my concept. While my concept Big Guy had a caring look, the model looked quite creepy and his overall igloo shape was lost. I agreed mostly but I just wish I could have been told earlier.

Post-Presentation #2

I decided to remake the model low-poly. I am still unsure if this was a good idea but I did manage to model and texture both characters and hopefully their quality is still to standard. Both characters come to only 2,600 tris which is a significant drop from 8,000 for one model. It was a challenge to try and capture the characters in a low-poly limit.

About 1,300 each

 As I was now very short on time I decided to skip Z-Brush, but i did try and go for a painterly texture sheet which was a bit different than I am used to. When I started the guy I had to find my feet for the style but I think by the time I had reached the girl I had found my footing. I do however need a lot more practice with skin as I think my faces have come off a little flat, and certain other elements such as the hair could also use some more practice.

The metalness, roughness and AO are all in 1 RBG texture sheet

Final Outcome

I am happy with the final characters but I am always wondering if the higher poly version would have been better. Due to the characters low-poly count I made the textures quite small too, which means that these characters probably wouldn't be viewed up close. I had also made some mistakes with the unwrapping, I should have had the facial features take up a lot more space over other parts of the meshes, like the legs or bags for example, then these characters might have been usable up close.

Week 10: A seaming pile of...

Once again I had a problem with my seams after doing the baking, but luckily I managed to figure part of the problem out. Despite the title, my frustration this week doesn't come from UV seams but from, ironically, the dichotomous views of the different lecturers/teachers on the course.

The week started off with me continuing on with the model, I had finished with the Z-Brush sculpt and was really happy with the outcome, and was now baking it. The seam problem I had with the turret popped up again, all the preparation work I had done with the normals didn't appear to have fixed anything. After some googling and tinkering I had noticed that 3DS Max was importing the bump files in with an increase in gamma, linked with the already high standard gamma it was making my normals look way over the top. This didn't fix the seaming completely but it did make it less noticeable.

It was just as I was finishing the sculpt when my work was looked over and given 2 thumbs up. There were some tweaks that were suggested, like relaxing the hand and adding in the other lantern that I knew I had to do, this was all going to be done after the baking. There were also a few other pointers that I agreed with and took them on board too. Everything was going well so I baked it, tweaked the model and then dropped it into engine to get ready for texturing.

We had a final presentation to do and this is what I had to show, as I had not got around to texturing or rigging the character yet. It was during the presentation that a completely different view point came in, mentioning that my model hadn't captured the concept. Parts of the critique I agreed with, like for example the face on the model is quite menacing, while the guy's face on the concept it actually pretty caring and rounded. At the end it was suggested that I restart the model with a lower poly count. I just wish this was mentioned earlier and not with a week left.

After working hard over the last few days I am now back were I left off, except this time with 2 characters and a much, much smaller tri count. The original tri count was 8,000, now there's only 2,600 between the both characters.

I took inspiration from the low-poly TF2 models

I have yet to decide if restarting completely was a good idea, I think when both characters are textured I'll be able to make a better decision, but the more I look at the pair together the more I prefer it; I can always finish the higher poly version during my spare time. For now I need to focus on texturing them both for Wednesday.

Week 9: Duos Continued

The last week has been a standard week, where I have been continuing on with the character duo project trying to get it completed for the deadline, only to find out that we've been given an extra week. This is great news overall but I wish we we're told sooner.

After the burning my original concepts received I picked up the pieces from the only thing that had got some positive crit, my bash kit, and I started from there. The pieces for the bash kit came from my original Pinterest/moodboards but this time I stuck things together with no real goal in mind; I just wanted to create some interesting silhouettes.

Bash kit into characters

When I had finally picked a silhouette pairing I began exploring what the characters could be and I settled on a feudal Japan theme, certain aspects of the silhouettes just made my mind think of that era. The final duo are a young girl and an older uncle/grandparent type character.  I also did not skip the colour iterations this time and I think they have already helped me create a stronger duo, but I do think that the process needs some working on.

Final iterations then the orthographics

The Duo!

I powered through the concepting stage over the weekend to try and keep on schedule as the deadline had not been pushed back yet, it would have been nicer to have been able to spend more time refining certain parts of it but I do think I have a stronger duo now than with my previous design. I'm still planning on getting this finished for the original deadline; the extra week I'll use to do some cool extra stuff that I'd love to do with the majority of my projects. I want to have a go at doing a final 2D piece of both characters, maybe another poster and if I can I might start modelling the girl.

At the moment I have just finished doing the base model of the big guy and I am preparing him for Z-Brush, I've taken all the precautions I can to try and stop the horrible seams my turret had but it will only be until after I'm done will I know if I have been successful.

By the end of the next week I'd like the model to be fully textured and in engine, but after the baking fiasco I had last time I'm not going to hold my breath. This assignment so far hasn't made me want to go fetal and I don't plan on it forcing me to either, the same time next week however I will have found out the outcome.

Week 8: Dichotomous Duos

The turret project is now behind me, for now at least, and a new assignment is upon us; the dichotomous duos. We need to design 2 characters from the same world/universe that belong with each other yet are different. My research board on Pinterest shows some good examples. After only a week though I have already hit a bump but I think I'm recovering from it well.

I started the project like any other way, with my notebook. This notebook has become a great way for me to just dump ideas into it as I tried to figure out some interesting combinations. As the randomization of the turret project worked so well for me I decided to throw some of that into this assignment as well by randomizing the theme.

From here I began creating the moodboards for my chosen idea, which was Creation & Destruction with an African theme. I originally planned on having them both set in the future too but I dropped that idea as i felt it was way too much. The final character duo was going to be a child soldier and a larger golem creature, that was like a remnant of a world shaper.

The moodboard expanded as I went through the design process

Next I started the idea generation. We were shown a way to generate ideas for characters by combining primitive shapes but this didn't feel like it was working for me so I decided to doodle some faces and after a few hours I had enough ideas to take them further to begin on the body. I continued until I had a final design for the first of my duo.

Designing the head alone was a mistake

Character 1: African Child Soldier

As I wasn't sure on how to generate ideas for my second character I picked interesting shapes from the moodboards and made another bashkit, just like for the turret. The results of this were overall much stronger than the other characters initial ideas and this was picked up on during our review presentation where our designs were roasted, the majority of us made the mistake of picking an idea then letting that guide our choices which lead to a lot of similar and bland ideas.

The simple silhouettes below were liked a lot more than my fancier final character above; I have now scrapped the war child idea and continued on with just making silhouettes for now. Hopefully this will give me a far more interesting duo in the long run, even if I have been set back a few days.

I think it is fantastic that our designs are being critiqued and we are being very strongly pushed to exit our comfort zones when it comes to our design choices, but there does seem to be a slight separation between some of the things we are being shown/taught and what is actually wanted. The start of the week we were being told that it's great to aim for a style like Dota 2 or Blizzard as it shows you can adapt yet by the end of the week it seemed like the message being said was the opposite. Next week I'll continue where I have left off here, with version 2 of my character duo and if things don't get set back any further I should have one of them modeled.

Week 7: Turret Post-Mortem

The turret project is almost complete, with the hand-in tomorrow and I am beginning to notice a common theme between projects now; things just are not going to plan. This week has left me once again feeling like I have taken a swift kick between the legs. I was also going to use this post to display all my awesome knowledge I gained in blueprint, sadly blueprint is harder than it looks and I struggled heavily with it but I will show some of the things I managed to figure out.

The Turret Design Process

The early stages of this project was fantastic for me, for once I had a great way to generate ideas and it was actually producing good results. I spent the time iterating on the design while also trying to incorporate my chosen splicing objects. Keeping the entire theme and style random was also a good way for me to explore areas of design that I would have never thought of.

Initial idea generation

I maybe could have spliced the objects into my design much earlier to get even better results but this just meant I had another iteration chance. I became stuck when I reached the final 3, I needed to start imagining these flat silhouettes as 3D objects but I couldn't do it, so I had to take a design into a 3D format so I could make a solid final decision. 

Final concept into orthographics

To keep on track I decided to start the model earlier than normal. I wanted to try and create a nice final 2D piece, showing what the turret would have looked like in its environment but I thought I would set that as an end goal incase anything went wrong. It's a good job I did. I also skipped the entire colouring process in a bid to get ahead with the modelling side, I got away with doing it this once I think; I won't be doing it again though.

3DS Max & Z-Brush

This is where things began to go wrong. I made the base model with a few hundred tris to spare and then I got into Z-Brush, all the earlier time saving was because I knew I wanted to use Z-Brush. I spent a few days in tutorials trying to figure everything out and I eventually left Z-Brush with some okay results for baking. The entire process is still a bit confusing though so I am going to note it down for future reference below.

Z-Brush sculpts

My Z-Brush Process

1. Create base model in Max, making sure the normals and smoothing groups are all okay.
2. Unwrap it.
3. Divide the base model into separate attachments, these will become sub-tools in Z-Brush which will make things easier.
4. Make a copy of the base model and make it high poly! Use edge-loops and turbosmooth (with the smoothing groups option).
5. Check the normals again (This could have been one of my big problems).
6. Export the model as an .obj and import it in Z-Brush.
7. Split the tool to create subtools, which makes things so much easier.
8. Before starting sculpting on a subtool Dynamesh it then divide, the number of divides depend on the object size (eg. a belt buckle won't need that many divides)
9. Once done, export the tool. Use the Decimation Master Zplugin if the tool has an insane amount of tris.
10. Import back into Max and bake!

Baking everything down is when it all broke. The bakes for whatever reason had huge, noticeable seams. I spent days trying to fix the problem, I must have baked each thing at least 15 times over trying little tweaks. Part of the problem I now realize was due to some poor unwrapping. I had the great idea to mirror as much as I could to increase texture space which lead to some very clear bump seams, so I redid the parts I could and hid a few others with geometry, like I should have from the beginning. The final model still has seams but the most noticeable ones are now hidden.

Texturing

My original schedule was now out the window, but I gave myself enough time to texture the model properly. Not doing a colour scheme earlier had slowed me down in places but overall things went smoothly until it came to dropping everything in engine.

Final 1024 texture sheets

The large grey empty gap was meant to be for the glass bottle, used with an alpha to get the bottle label to show up. When in engine however, the translucent option broke my model by making all polygons to appear flipped or something similar. I made the decision to use a multi-sub, with a glass material being used for the second slot, this meant that my label and the bottle holder wouldn't be visible but I thought it looked much better than an ugly grey block.

I also placed the metalness and emissive textures in the RGB space to save resources, but looking at it now I could have thrown the roughness in there too.

UE4 Screengrabs

UE4 Material Help

Glass

Quick bump map increase

Simple Glow

Finishing Off

Originally I wanted a couple of days to mess around with Blueprint but in the end I just had one, where I manged to get both barrels to emit the firing particle and eject a cartridge. I couldn't figure out how to get the second barrel to actually fire or how to make the small crank rotate. Spending days staring at blueprint and trying to find tutorials has left me quite exhausted. Maybe just before hand-in I can quickly grab some help.

We had also been tasked with creating a promotional poster for our turret. I was a little nervous in creating this as I didn't know how much of my graphic design knowledge has actually stuck around in my head but I am really happy with the final result. I tried to keep with the theme of late 19th Century Russian by injecting a small amount of constructivism into the filigree.

Once again I am feeling a little underwhelmed with the end results after 3 weeks. Maybe I need an ambition check once and a while to scale myself back, I am pleased with how the turret looks but when parts of it don't work how I want it has left me dissatisfied. I also had to heavily cut back on the 2D side which is something I am not pleased about, I still might try and paint the turret in an environment, even if it doesn't get marked it will still be great practice. As for the next project, bring it on! What else could go wrong?!

Week 6: Z-Brush

I mentioned last week that I wanted to get into Z-Brush, well, the last few days that is exactly what I have done and after a few stressful hours I have finally managed to create something that I am happy with.

The week had me start with finishing my base model from the week before and testing it in engine. I wanted to make sure it worked before I moved any further and thankfully it works fine with the skeleton. I will need to make adjustments though, like making it fire from both barrels. I have then spent the rest of the week fighting inside Z-Brush as I tried to get use to the unique way the program handles itself.

Final turret design

Turret base mesh - 5,000 tris

I found that many of the tutorials for Z-Brush like to focus on sculpting a super-awesome rock dinosaur robot and the process behind that, which is great for when I'm ready to tackle something like that but at the moment is pretty useless. Simpler, beginner tutorials appear to be pretty difficult to find. This just slowed me down, as I began smashing buttons and trying things out myself and eventually after a number of failed attempts I began to have positive results.

This might actually look like I know what I am doing in Z-Brush

There were plenty of tutorials videos on  the actual baking down process though, which was great. Below is the bump map of my various bakes. The massive gaps are because I didn't do high poly bakes for every part of my turret, these will be filled in with bumps from textures. Being new to Z-Brush I wanted to keep things a little simple.

Next week is when I plan on getting the turret working in engine properly and I will be making a bigger blog post about it, where I'll be noting down everything I pick up on to make it work. This will hopefully serve as a good resource for me if I ever need to reuse something from this project. Below I have linked the final look of the Viking room we created, as a group, during the first week back. I have been meaning to post it up but every time I have forgotten about it! I concepted the bench, modeled the hog's head, unwrapped the shield and textured the barrel! I also made the rug as an extra. For now that is me done, I need a break from this turret and more importantly Z-Brush!

Week 5: Sentry Turrets.

The third project is, as you can probably guess by the title, all about sentry turrets and we've been working on it for roughly a week now. I can't tell if all the group work has made me really grateful about a solo assignment or if this project is just ticking all the right boxes but I've so far been enjoying the Sentry Turret a lot!

A sentry turret is basically a self-operated, stationary gun that will normally be protecting a corridor or some high value objective [1]

This assignments final outcome is a sentry turret that tracks the player in UE4, and fires. We've been given a skeleton base to rig it with but apart from that we have complete creative freedom, which sounds great but when you get down to putting something on paper it suddenly becomes quite daunting. In one of our lectures we were shown a way to overcome this hurdle with a few randomizing techniques and I put them to good use because I knew I'd fall into the same old trap of doing some post-apocalyptic snore turret.

After some random word generation and a dice roll for the time period a theme was chosen, a late 19th Century Russian poison vodka turret. I found this particularly interesting because it bypasses the now stereotypical view of a Soviet Russia. Having an idea I started working with some moodboards.

For an extra challenge we also had to splice an interesting object into our turrets, last time I did this I chose a mini-wheelie bin which made it much harder than it needed to be, so this time I went on the hunt for an actual interesting object! I soon found 2 objects, one was a toy train from a kids meal and the other was... something that caught my eye while I was looking around Leicester market... I'm sure you'll know what it is.

The toy train and "smoking appliance"

The entire concept design process is still something that I am learning but with a helpful lecture on creating silhouettes I had a blast during the early stages of design, which has so far made everything else more enjoyable too. I started with a bashkit to build up some silhouettes and iterated from there.

The only part that I struggled with was towards the end of the process when it come to picking a final design. My head couldn't fully figure how these  flat silhouettes would work in a 3D space, which is why I started adding values then had to stop to build this turret in 3D. To help speed up the perspective (and make sure it was correct) I built a quick mesh in 3DS Max. This did take a few attempts but the end results I think are very strong and I am now in the process of combining the different aspects that I like into a final design, and I should then be able to begin modelling.

I managed to look into Z-Brush and it's clear that the program will require more than a weekend to get my head around, but it would really help with this project, especially with all the filigree. Forza Horizon 2 is still keeping me busy, to the point that I've only bought 1 game in the Steam Halloween sale, and I've also done another update to the blogs background, but I can't tell if it's effecting my eyes while reading  so I might need to have a rethink about it.


Notes & References
[1] Turrets from Team Fortress 2 & Portal, both created by Valve.

Week 4: And Relax.

The film room project has, at last, come to an end. In my last blog post I mentioned stress doesn't effect me much but I can say that last weeks work almost had me reaching high stress levels due to me trying to finish the work on time.

The week started off with me texturing the objects for the film room, it was very clear at this stage that the room was not going to be finished so the group prioritized the most prominent objects. I did the couch, chair and foot stall and to be time efficient I used one texture sheet for them all, unwrapping the objects became the slowest part of the process. The lack of time also meant my UV's couldn't be packed as tightly as I wanted.

I really wanted to get the room finished, putting in so much effort for the final thing to be uncomplete felt terrible. The final presentation was also looming in the back of my head because the room wasn't complete, yet thankfully the teachers understood fully the situation our group had got it self into and was overall impressed by the end result, other people in the course were also very supportive of the final room, despite it not being finished.

The final room

The final, spliced with the original shot

The final presentation gave our group plenty of time to review what went wrong, and it's clear that we had spent way too much time concepting, especially for objects that were in plain sight. This then had large repercussions for everything else, to the point that we were messing around in the engine the night before, trying to match the lighting as best we could. Our overall time plan was also far too restricting, with little room for error and not enough time for engine work. Picking a simpler room might have helped too, for 3 people the living room might have been a little too object heavy.

The next project starts on Monday and I will hopefully be able to get into Z-Brush this time around, as I think that would be a great way to push my models further. To help recover from the last week I also picked up Forza Horizon 2, which so far has been amazing, even if I haven't played it as much as I would've liked. Overall though the Film Room project has been quite the learning adventure, just not in the ways I thought it would be.