Ryan Jessop Final Major Project - ARC Troopers

Ryan_JJ

Hello all,

This is my progression thread where I will be sharing the work of my final major project at University. I study Masters in Games Art & Design with a previous background in Architecture. This project is aimed to showcase a concise variety of technical skills based on a game concept I developed within the Star Wars Clone Wars time period. 

Link to my Pre-Production pack: https://www.linkedin.com/posts/ryan-jessop_ryan-jessop-final-major-project-pre-production-activity-7196839091974406146-vfG3?utm_source=share&utm_medium=member_desktop

My submission is August 21st 2024, but I intend to keep on pushing this project further afterwards in my spare time.

Hope you all enjoy!

Ryan_JJ

A sample of my Pre-Production planning to give you an idea of my concept and what I am looking to create.

Ryan_JJ

.

Ryan_JJ

.

Ryan_JJ

Ryan_JJ

Ryan_JJ

Ryan_JJ

Ryan_JJ

Ryan_JJ

Hey all who are reading, just thought I'd share some work for my first tasks this week.

From yesterday (Monday) I have been working on Modelling the LAAT Gunship. I have used this image reference to help get the proportions right. This modelling & texture task is designed (by me) to get myself warmed up and back into the swing of things.

This is what I acheived by this morning:



I took it to my class tutorial today and was given commendation for it's accuracy and looks. But I was slated by the dense topology as this was suppsoed to be the Low-Poly... (I got carried away).

Since I've been home, it's taken me barely any time at all tp produce a new model:

This is far more efficient as because I'd already gotten used to the geometry, curves, bends etc, I knew where I to put my loop cuts more efficiently. This is with 1 level of sub-division using the 'subdivison' modifier.

More to come soon

Ryan_JJ

The wings are on! While very simple to do, it drastically helps visualise the scale and proportions

Ryan_JJ

I'd been dreading this part of the build and I was in split minds on the method to model this.

Do I refine the edge by adding more loop cuts and make the model higher poly?
Or do I do a Boolean difference and suffer the affects of N-gons?

I asked my class mates on our discord (a really brilliant group and I'm very lucking to be on this journey with them) and a few pro's told me I could just mark the edges as a crease which would stop the edge shading as smooth curve and shade as a sharp edges instead. 

(I'm still learning modelling techniques if you can't tell...)

Ryan_JJ

This is the result and I think it looks pretty good! Easy to refine as well if need be...

Michael Knubben

Bad news, and I assume your teachers will tell you the same: This still isn't 'lowpoly'. Anything you use a Subdivision modifier on is a subdivision mesh.

I'd block it out in a very simplistic way, and then bevel (with a modifier, preferably) any edges that need it. You can use your highpoly mesh as reference, but a lowpoly never has holding edges intended for a subdivision process.

Note that, even for a highpoly, those edges going all along the entire mesh will make it harder to work with too, you'll want to terminate before.

Ryan_JJ

Understood but it's suitable for use for now. This is my first time modelling any type of vehicle so it wont be anywhere near up to industry standard xD. Definitely a huge improvement over my earlier models!

Alex_J

hey, really cool project, awesome to see the learning process

just adding on to what michael has said, a few tips to make modeling go a little easier: 

you only need edges to define the silhouette. It can help to use flat shading in the viewport so that the only thing you can see is the silhouette. Orbit around the model and so long as silhouette is good, you have edges where you need them. Then you can delete any edges which aren't holding the silhouette. While working on the big shapes, it's okay to have it not quite as smooth as the reference. So long as you have the extremities defined, you can easily add edge loops in and have the computer interopolate to smooth it out later.
The benefit of working this way is that you can very easily adjust the shape because only a few vertices to move. So if you realize the proportions are a little off, you can fix that easily.

It helps a lot, especially as a beginner, to limit the size of your problem space. So rather than trying to solve two problems at the same time, only focus on one. So I'd suggest to focus only on the shapes of the models and not worry about shading yet. If you have a low poly model it is very quick and painless to modify in the future, should you need to add an edge here or there or change an edges normals (smoothing). 

One of the biggest things that will make topology problems go away is to construct the thing in pieces rather than one seamless mesh. Usually the same way it may be constructed in real life. In real life you'd imagine there is a chassis and then panels on top of that. It's not one contiguous panel.

For example, the ball socket joint panel on the nose which houses the weapon - this doesn't need to be connected to the main body panels. You can see a seam in the reference, suggesting it's a separate piece. 
The benefit you gain if you model it separate is that you can have different topology and not have to think about how to join it. 

If you are eager to see what the final thing may look like and get a better handle on the entire process, it may be worthwhile to take just a single component of the entire ship - like say the weapon turret - and take that to completion in isolation. Then everything you learned will improve the process for finishing the rest of the ship.

Ryan_JJ

Thank you very much for your comments Alex (and by extension, Michael). I totally understand what you mean. The reality is, this is my first time modelling something like this with such curvature and so as I began modelling the main body of it, I made some destructive mistakes (i.e adding too many loop cuts where they were not needed). As I progressed through, I started to realise that I could have done different parts more efficiently. However, as I am against time for this project, I cannot go back and redo it at this stage.

What I intend to do is take your comments and tips forward for future modelling parts of this project. And if I have time at the end, I'll come back and re do this.

I have found that redoing the modelling becomes quicker and quicker and becomes more and more efficient each time so (fingers crossed) the next time I do this and come back to it, it wall be a quick process.

re: the ball turrets, yes I agree and I have modelled them as completely different meshes, same as the wings and the rockets/thrusters on top.

But the main body chassis I think really has to be 1 solid piece with added details later down the line.

For the purpose of my assignment, this will main serve as a static prop and will be put into UE5. So worst case scenario, I'll rely on nanite to help curb performance tanking in engine. 

Ryan_JJ

I really needed to get on with other bits of the project but I really couldnt help myself, I had to see if I could do this mesh better (yet again) using the ways you mentioned. Here's what I've done in about 10 mins!

Ryan_JJ

I'm stunned by how quick I was able to do this considering the amoujtn of time I had spent on the previous version. Lesson has definitly been learnt!

However, in the interest of progress and my time planning, I will be taking the previous model version forward within development. While this new retopped version is looking great, it needs refinement whihc I don't quite have the time for just yet. As mentioned previously, I will come back to this when I have a spare window of time.

Thaks again for your input guys! Hope to speak with you both again later into production on other bits n bobs
 

Ryan_JJ

Ryan_JJ

Going against what I had said before, I decided to jsut go for it and retop the main body of the model again. I've kept it pretty efficent this time around (I hope you all agree) 


The thrusters/rockets on top i'll redo. But what I have found interesting is doing the boolean difference modifier operation for the main side doors adds a lot more faces...



So I'm considering  not doing a boolean difference operation and going through the pain of cut out the doors with the knife tool. However I'm not sure if I'll be able to get the doors to match up to the curvature and geometry perfectly to the rest of the ship if I do so.

decisions decisons....

Ruz

I would stop with the long thin strips, go for more square poly/even distribution. think about the silhouette and try to be minimal in capturing the

look. think about the wasted loops where nothing is defined adding extra lengthways loops

Ryan_JJ

Ruz said:

I would stop with the long thin strips, go for more square poly/even distribution. think about the silhouette and try to be minimal in capturing the

look. think about the wasted loops where nothing is defined adding extra lengthways loops

Thank you for your advice, Ruz. I have tried to get it as 'blocky' and basic as I can so far. But I think I'm just not quite there with my skills to be making stellar efficent models. I am pushing myself to be better, and with everyones input here, I have improved it a decent amount already which is great. 

I am curious though as to how many faces/tris to ball park for the model? I took some inspiration from your CG Trader vehicles which are super efficent. The main body/chasis is the main focus currently as it makes up the majoirity of the ship, and its curvature is tricky to get right as some of the changes in shape are very subtle (hence why it may appear im added unnecessary details nad loop cuts)

With the sub-div modifier off, the main body is 370 faces / 727 tris but obviously looks quite blocky and sharp.

With the sub-div modifier on the main body is 2,715 faces / 5644 tris. 

With the sub-div modifier and the solidify modifier on the main body is 5,430 faces / 11,276 tris. 

Ryan_JJ

Quick WIP update...

I've remodel this thing about 5...maybe 6 times now? Obviously this isnt sustainable but all part of the learning process I guess. I've managed to get the doors cut-out WITHOUT doing a boolean difference which I did not think I'd be able to achieve. But over the past week I've become gradaully more skilled (still not a pro but better than beginner at least?) so I've been able to cut out the doors instead rather than do a boolean differecne and have to tidy up a shit load of n-gons...

Here's where I'm at now...


34k tris for low-poly. This is with a shelled interior, but no details but in yet (over head handle bars, door hinges, greeblings etc) And I still need to model the cockpit.
Doors look a bit more dense in poly's compared to the rest of the ship, but this was necessary to get the windows cut out whilst retaining the exact form of the ship. And my tri count is pretty efficent I feel like so I can afford this?
Nose of the ship is more dense also. Again, this was sort of necessary to get the front ball cannon sockets to recess as I wanted.

Hoping to get this all finsihed up and UV unwrapped tonight to take to class tomorrow. 

I am a few days behind schedule as tomorrow I was suppsoed to be finished with texturing. 

I'll be making some adjustments to my time planning later in the week...

High poly (2 sub-div levels currently).. 90k tris

Ryan_JJ

Another update...

Very closed to finishing all exterior modelling. I will not be modelling interior ar this stage. But If I have time at the end, I will do so...

Anyways...

Low Poly model is 34k tris


High Poly model (2 sub divs. Will most likely do 3 or 4 when I export for baking) is  280k tris.


Hopefully this is good enough to work with and bakes well! 

Just need to model some very small details on top. A little mechanism that loads rockets into the launchers.

After that, my tutor recommended I use UDIM's for this to UV unwrap and texture. Will be looking into this tonight.

Alex_J

is this is intended for games? 

If so, I might ask around a bit about UDIMs. I thought that was something only used in film (or at least not games). To finish the assignment probably best to just do whatever teacher wants but if trying to get a job in games it's going to be good to know what workflows work and won't for games. If you will render this in a game engine you probably ought to do a test to double check that your UV/UDIM workflow will work in the engine.

Ryan_JJ

Alex_J said:

is this is intended for games? 

If so, I might ask around a bit about UDIMs. I thought that was something only used in film (or at least not games). To finish the assignment probably best to just do whatever teacher wants but if trying to get a job in games it's going to be good to know what workflows work and won't for games. If you will render this in a game engine you probably ought to do a test to double check that your UV/UDIM workflow will work in the engine.

Good point. So the use is sort of both... I wanted to do the Low/High Poly workflow as a portfolio piece to prove I can do that. The use of the ship will be seen up-close in game but may also be used in cinematic/show reel content. And I was recommended to follow a UDIM workflow by my tutor. I wanted to follow-up on that before commiting but he wasn't in today unfortunatly. I'll ask some class mates though and gauge opinions before fully committing.

Ryan_JJ

Had my head down the past few days but am happy to share some progress... 

High Poly mesh has been baked onto Low Poly mesh pretty well. I have decided to use a single UV tile and not go for the UDIM route.

Ryan_JJ

Current bake with a quick smart material slapped on:



I was hoping to have this all textured by now but I ended up redoing the modelling (a lot) more than I first anticipated. All part of the learning process at the end of the day but I will definitly need to make up for lost time somewhere.

I've also taken a look into Anchor points, decals, stamps, stenciles, overlay filters and decals to better my Substance knowledge and workflow. I feel a bit more better equiped in going into texturing this now. But I am a bit stuck on how I will do the 'panelling' look on the main body and wings as seen in the image below

Ryan_JJ

Quick update. I'm still a bit stumped on how I will make the lines/panelling look in the image above ^^

But I have pressed onto texturing regardless to keep the ball rolling...

Was a lot of in and out of baking and remodelling to get the High poly mesh as close as possible to the low poly mesh. I now have in a state which I am happy with. 

I have also taken some time to learn about anchor points and have used them to create some hull damage. Below is an example and is subject to change...



What I am trying to acheive here is a texture/material build up. Bottom layer is a sort of sci-fi circuit board texture. In the middle is a metal material and above that is the main paint layer. 

What I am trying to aim for is to paint on a paint layer (with anchor point) and to then manually pain in areas where the ship has sustained damage. Some worse areas in others. Some as just scrathes where the paint has come off and other areas are worse which destroy the metal and expose the mechanical componments and circuitry inside. 

I used these tutorials to acheive this:
https://www.youtube.com/watch?v=q9mLZQKxPSo
https://www.youtube.com/watch?v=SvavIvIm9sE

As I say, still playing about with it, and the chosen materials and textures can be easily replaced and subbed out without losing my damage placement.

If anyone has any suggestions on the panelling PLEASE speak up!!

Thanks everyone

Ryan_JJ

Closer look:

Alex_J

i'm not much of a hard surface modeler, but my first hunch for the panels might be to model them and bake them as floaters. I think that would just be simpler to do rather than try to paint it in texturing. And if you bake it you can get curvature maps and so on which can give you more flexibility in texturing. 
you might do a forum search for "panels"  and similar terms, probably a lot of historical examples floating around might give some ideas as well

Ryan_JJ

Alex_J said:

i'm not much of a hard surface modeler, but my first hunch for the panels might be to model them and bake them as floaters. I think that would just be simpler to do rather than try to paint it in texturing. And if you bake it you can get curvature maps and so on which can give you more flexibility in texturing. 
you might do a forum search for "panels"  and similar terms, probably a lot of historical examples floating around might give some ideas as well

Thanks Alex, I'll start searching the forums now. 

Could you also elaborate on what you mena by 'bake them as floaters' ?

Alex_J

On link below page, press ctrl+F and then search "Panel", at least one example there

Subdivision Surface Modeling - polycount
also check google as well if needing some more step by step for the baking process, looks like one good tutorial here: 
ArtStation - Understanding High Poly Floaters for Baking Details

Ryan_JJ

Thanks again for everyones help[ thus far. I looked into Floater island baking and I gave it ago with mixed results.

I'd like some critique on my texturing so far. I am putting this down now for a bit to get on with other parts of the project. But I'd like to be able to come back to this at a later date with a clear idea of what needs to be improved and how.

I know it needs a lot of work still. But I've been looking at this so much the past few days its become quite tricky to point out and figure out where the weak bits are. They might be blindingly obvious to you. But I've just gotten used to them and grown acustom to them. Hence why I would greatly appreciate more eyes on this and opinions. Thanks!

Ruz

panels are loooking ok in that last image. lots of different ways to do panels tbh

you can stamp normal map details in Substance painter( the panel on the very right was stamped on) and some people even do the panels in susbtance

I was messing with my bus , adding a few details, but I have only just started learning this recently

Obviously this relies on having good texel density or the little rivets will look really low res

Notice my lines came out a bit twisted, thought they look fine in substance

also re the image of the girl , maybe feed that through Kria ai. I tend to do that on most of my logos, but that's up to you

Probably best not to get too bogged down in detail at this stage, maye focus on getting the livery colours/design looking nice

you didn't seem to add any height detials to the dirt either, maybe look at that too

https://www.youtube.com/watch?v=JfDNyP0nDjI&t=2s

https://www.youtube.com/watch?v=y_VMS9yl38Y

Ryan_JJ

Hello to all who are following. 

Had my head down the past couple weeks. 

Again, thank you for everyone's support nad feedback thus far!

The LAAT Gunship has been paused for now so I can get on with the all the other bits of the project. Though I do hope to come back to the Gunship toward the end to improve, maybe remodel some of it...

Anyhow, I have been creating some of the environment assets from my original concept art:


https://www.artstation.com/artwork/xDkVQY


Midjourney generated

Ryan_JJ

Computer Terminal:

I've made 2 terminals for this as there will most lilekly be a few of these in the environment. So having a variety will help break up the repetitiveness. 

Concept art & source images:


Current model:

Ryan_JJ

Wall Planter:

These will be quite prominant in the interior of the lab but nothing too complicated...

Concept art generated:


Current model:

Ryan_JJ

Floor Planter:

As with the wall planter, these will be a focal point in the environment. I wanted modularity and flexibility to adapt the sizes as needed. For this I used splines and a profile outline mesh so that I can easily change and adjust where necessary. 

Concept art:


Current model:

Ryan_JJ

Plant containers/crates:

Few of these will be scattered around, and they'll all have plants inside them. I'm hoping to make my own foliage with SpeedTree but that's for way later down the line...

Planter 1....

Concept Image:


Current model & UV's

Ryan_JJ

Planter 2:

Concept image:


Current model & UVs (not sure what's going on with the odd shading...but this is WIP) :

Ryan_JJ

Planter 3:

Concept art:


Current model & UVs:

Ryan_JJ

Terrarium Pod & Platform:

Original Concept Art (Reminder) :
https://www.artstation.com/artwork/xDkVQY



Current Model(s) :

Ryan_JJ

That's pretty much all my work from the past couple weeks... This has been a bit of an overdue work dump which hasn't be laid out very preofessionally but I wanted to share these with everyone asap.

Had a few back and forths with my tutor on my UV unwrapping of the plant pots/storage crates. Previously the UV's were (apparently) awful so I've redone them with some guidance. Bit more TLC needed on the onnes which are done but they're heading in the right direction I think... I recognise that the orientation of some of UVs needs to be corrected and the packing needs to be a lot better as there is far too much wasted space. 

Over the next few days I'm going to focus on UV unwrapping the rest of these assets so I can spend next week doing the textures. 

below is a screenshot of the assets I originally conceptualised and broke dowrn for me to make. I have done most of them but I may use an asset pack from ArtStation for the greebling if I feel I don't have time to create my own.

Ryan_JJ

Hi everyone, just some more updates incoming...

Firstly, here are some textured assets...



For this I set up a Metal smart material wiht some tutorials online and tried to make a 'steamed glass' material for the glass pod in the center. I think it works pretty well but I have made refinements to the material in later assets as I'll share soon...

The metal needs tweaking I think. Not sure if I like the roughness of it. The curvature mask fro the edge ware needs dialling way back on the side clamp parts which connect the top and bottom plates.

Source image generated with MidJourney:


I have of course taken some creative liberties with this and am only using the generated image as a guide.

In order to get the steamed/condesnation effect, I followed a tutorial online and made some tweaks. The smart material is set up to work just on roughness with grunge maps masked over. A paint layer is also added to the mask so that I can paint on condensation with a physical paint brush. This makes it so the water droops run down the face of the glass naturally. The premise of this material I learned from this tutorial.

Ryan_JJ

Another one of these plant containers... This time I have dialled it back a bit on the edde ware and weathering. The glass material also appears a little more realistic as I tweaked by customer 'Physical Paint' brush in Substance Painter. I also gave the roughness a very slight height bump so the condesnsation as a fine amount of light catching the sides of the water drop marks to help make it more realistic.





Source image generated with MidJourney:

Ryan_JJ

One thing which has caused me some grief is getting the texel density correct. Both of these assets are around 6 or 7 px/cm. I wanted it to be somewhere between 10-15 px/cm. If this project were to be realised into the game I have envisioned, then it will most likely be a third person shooter. From what I read, texel density ought to be around the 10-15 px/cm range.

I have used Tthe on-board Blender UV packer and free UV-Packer add-on but they both pack to similar results.


I have managed to get RizomUV student license but need to figure out how to use it correctly. For now, in the interested of time (deadline August 21st) I was recommended to just up the UV resolution to 4k (currently 2k). But going forward I should look into overlapping and rotating UV's or replicated parts of the mesh.

Ryan_JJ

I have also done some work in UE5 for ym blopckout. previously I had it all in Blender just becasue I found it easier to manipulate and duplicate/array objects as I wanted. But now I need to set up my golden path and grand vista shots.

Block-out is rough but here it is...


Original Concept Art (Reminder) :
https://www.artstation.com/artwork/xDkVQY


I really wish Unreal Engine had intuative array tools such as Blender. I struggled using the modelling tools to create 'patterns' for my meshes adn instance them as I wanted them. I am hoping to play with this all of tomorrow (sunday 21st) to get evertything set up correctly with the lighting as I want to be also. Then can spend the following week finishing off textures and bring everything over to engine.

Any block-out snazzing tips and tricks are much appreciated!!!

As a base stgarting point, I will be using Karen Stanley's beginners guide: https://kazperstan.gumroad.com/l/tMKaf?layout=profile

Ryan_JJ

Some developed camera shots to set up my scene. All WIP but gradually getting there. I hit a bit of a creative block so I've stepped away for a few hrs and come back to it (though it is quite late in the day now) I am now going to sketch over this in Procreate as I find connecting my had to the page allows for my ideas and creative decisions to flow better than having my hands on a keyboard and mouse.... maybe im old fashioned but that approach was drilled into us Architecture students from day 1

Ryan_JJ

Few more updates of the project...

I have made my first custom shrub in SpeedTree. Quite simple but im happy with it regardless. Especially the emmissive parts  

My process for this was...

1- Use Midjourney to create concepot art of sci fi plants.


2- Reverse engineer the concept art and draw/paint the leafs, I used Procreate for this.


3- Send the drawn leafs back into Midjourney with a generation prompt such as " photorealistic detailed sci fi blue leaf Canon R6 Mark II, 35 mm lens  --iw 2 --style raw" This helps making the leafs go from a stylised look to a realistic loof. 


4- Compile the images in Photoshop to create a texture atlas, normal maps and alphas.
You now have all the textures you need to make a simple sci-fi, completley made up shrub within SpeedTree  

It's not the most perfectly photorealistic workflow. But it's pretty easy and yields decent results...