Jump to content
 

The non-railway and non-modelling social zone. Please ensure forum rules are adhered to in this area too!

It's modelling Jim, but not as we know it..


froobyone

Recommended Posts

Thought I'd share the last 3D project I did.

 

When creating 3D illustrations like these, I find I use a lot of techniques developed in real world modelling. Especially when it comes to texturing and weathering the pieces.

 

But the best thing about 3D modelling compared to real world modelling is, if you make a mistake, you can restore to an earlier version. Which, as I'm prone to mistakes, can only be a good thing.

 

Anyway, here's the pic, I include it as it has a bit of railway in it. :)

 

RushDeneMiningCo.jpg

Link to post
Share on other sites

OK I'll bite:

 

if you could just hit Print and have a painted 3D scene come out of the printer...

 

Fooled me at first sight, but then when you look closer, there's something odd about the wood textures. Can't quite put my finger on it. The lighting's pretty convincing.

What software is this?

Link to post
Share on other sites

I can see a market here for some enterprising individual to produce generic backscenes to add depth to layouts

 

It's funny you should say that, I did consider using the technique to create a backscene for Hullbridge, but never got around to it.

 

Fooled me at first sight, but then when you look closer, there's something odd about the wood textures. Can't quite put my finger on it. The lighting's pretty convincing.

What software is this?

 

I'd be interested to know if you can put your finger on it. Is it the main peice of wood in the foreground or the water tower? If it's the front peice, then I undestand what you mean. Even though it's a 4k texture file (with 190 layers o.O), It's still not high enough resolution for an up close and personal.

 

Here's a pic of the tower before the rest of the scene was built, it shows off the wood on the tower much better I think.

 

RushDeneMining4.jpg

 

And finally another iteration, that uses DoF to hide the texture a little bit on the foreground.

 

RushDeanMiningCoBeta1.jpg

 

It's all smoke and mirrors.

 

The software is 3ds max 2011. I still wake up in the night screaming at the thought I spent £2700 on a piece of software!

Link to post
Share on other sites

I'd be interested to know if you can put your finger on it. Is it the main peice of wood in the foreground or the water tower? If it's the front peice, then I undestand what you mean.

 

I think it is that the wood has, if anything, too consistent a matte finish. Real wood as harder patches that, while not necessarily reflective, are almost shiny. This is particularly true of creosoted wood. The attached photo has too much shadow to really show it properly, but it was handy.

 

post-206-051419500 1286971446_thumb.jpg

 

Adrian

Link to post
Share on other sites

I think it is that the wood has, if anything, too consistent a matte finish. Real wood as harder patches that, while not necessarily reflective, are almost shiny. This is particularly true of creosoted wood.

 

Adrian

 

That is spot on Adrian. I didn't add a specularity map to the wood due to the fact that the main texture map was so complicated. The spec map that I did try, didn't look right at all, so rather than spend another day trying to get it right (it took two days to build the map for the main tower wood), I opted to leave it off.

 

But kudos to you for spotting it. You do realise though, that I'm going to have to go back to the model now and do the spec map again. -_- lol

 

I did have two more views of the 08.

 

08.jpg

 

 

0820WIP2032.jpg

 

 

You may notice inside the cab, the addition of the dials. Yeah, you're right, no one sees them.

 

I also have to say, for the rivet counters (joke) that each rivet is a seperate object and as far as I can tell, they almost excatly match the real thing. See if you can spot one that's seemingly free floating. It's ok, I obvioulsy missed it too. :)

 

As for the leaf springs, don't get me started...

Link to post
Share on other sites

That is spot on Adrian. I didn't add a specularity map to the wood due to the fact that the main texture map was so complicated. The spec map that I did try, didn't look right at all, so rather than spend another day trying to get it right (it took two days to build the map for the main tower wood), I opted to leave it off.

 

But kudos to you for spotting it. You do realise though, that I'm going to have to go back to the model now and do the spec map again. -_- lol

 

 

I added a photo to my post that may (or may not) help.

 

Adrian

Link to post
Share on other sites

I'd be interested to know if you can put your finger on it.

...

I still wake up in the night screaming at the thought I spent £2700 on a piece of software!

If it's any consolation on the spending, I must have spent nearly that much on upgrading Adobe CS over the years, always for very minor improvements, and just for client compatibility.

 

 

Trying to put my finger on it...

 

On the big timber in the foreground, the vertical bit nearest the "camera" is sharp, but the horizontal bit seems a bit fuzzy/out of focus, which jars with the sharp details further into the scene. The revised version with depth of field fixes that.

Also the open grain on the vertical bit of that big timber seems a bit "flat" in the open parts, as if the darker bits should be darker still.

The very open grain suggests a very very old and weathered bit of wood, yet its edges are quite sharp.

 

On the water tower, maybe there's too much texture compared to what I was expecting.

 

Having said that, it's amazing work though. I'm still astonished by just how far the software has come, and what it can do in the right hands.

How many hours to go from nothing to what you posted?

Link to post
Share on other sites

I also have to say, for the rivet counters (joke) that each rivet is a seperate object and as far as I can tell, they almost excatly match the real thing. See if you can spot one that's seemingly free floating. It's ok, I obvioulsy missed it too. :)

 

Topmost rivet on the nose seems to be floating in air (based on its shadow).

 

Adrian

Link to post
Share on other sites

Topmost rivet on the nose seems to be floating in air (based on its shadow).

 

Adrian

 

Yep, I agree with Adrian, its alignment with the rest of the column of rivets seems awry, compare pics 1 (post #5 ) and 2 (post #9), it seems to have "jumped ". In pic 3 it seems to have disappeared or do I detect it in the yellow painted area?

Exits nit-picking mode!

Link to post
Share on other sites

To me what is wrong with the grain effect on the timber is showing that it is going across the timber not the end of the grain as it should be as this will be a fault of the program as I have seen this on another program so I looked hear first.

Link to post
Share on other sites

To me what is wrong with the grain effect on the timber is showing that it is going across the timber not the end of the grain as it should be as this will be a fault of the program as I have seen this on another program so I looked hear first.

Whereabouts do you mean? If you mean the horizontal banding on the end grain of the big baulks holding up the water tanks, I took them to be lichen.

Link to post
Share on other sites

Just to wind you up slightly my daughter was given 3dsmax as part of her course.

 

Most likely this will be the student form which will be a short time license for the length of the course she is on.

 

Whereabouts do you mean? If you mean the horizontal banding on the end grain of the big baulks holding up the water tanks, I took them to be lichen.

 

Yes this is where I mean and to me it doesn't look right against the rest of the timber work on it.

Link to post
Share on other sites

What marvellous eyes you all have. It's these things that make you great modellers.

 

Yes, the end grain on the tower timbers are incorrect. It's actually not a fault of the software, but the fault of the operator.

 

I'll try to explain the process so you can understand what went wrong. Each seperate piece of the tower, every plank and every support has to be "UV Unwrapped" It is in effect laid flat and then displayed as a jpg, 2d object. So every piece had a section of 4 sides and then the end peices are seperated off. This was done on every peice, of which there are many. Then every flattened, or "unwrapped" element was placed into a 4096px jpg file.

 

You have to size each element to maximise the screen space at the same time as fitting everything in. Things that are large in the forground or just large in general, need more screen space than things that are small. The bigger an element is, the more texture detail it can hold.

 

Now, my end peices ended up getting mixed up with some other end peices, that had been sized much smaller, so the area of the texture map that I thought was paiting the end peices, were in fact painting smaller insignificant parts of the model.

 

Here's a downsized version of the final texture file.

 

watertowernewdiffuse.jpg

 

You may be able to spot some of the main parts of the model in here.

 

Up in the top left of the image, is where all the end peices went. But 4 of the really small ones, were in fact the large end peices. Hence they didn't recieve the right texture.

 

To fix the problem, I would have had to UV Unwrap every single peice again, a process which took me 18 hours, split between two days and also rebuild the texture file from scratch, (as I've mentioned, there was upwards of 190 layers and a total of two days screen time) because no two peices would have been in excatly the same space as the first unwrapped file. You paint over the UV's as a background, like using tracing paper. The software then knows where each bit of paint goes, based on the UV co-ordinates held for each peice of the model.

 

So that's why the end peices are wrong, well spotted. As this project was just for my own edification, I decided not to add 4 more days to the build.

 

To answer the question how long it took from start to finish?

 

Tower:

Two days building the geometry.

Two days UV Unwrapping.

Two Days Texturing.

 

Mine cart

Two days geomoetry.

1 Day Unwrapping.

1 Day Texturing.

 

Rusted lamp.

2 days all in. Although the design of the sides is very complex, once it is done it's simply copied and rotated around the centre point.

 

Environment

10 minutes building up geometry of the landscape.

Two weeks getting the trees/distance mist/lighting. It went through a lot of iterations before I was partly happy.

 

There was another pretty good reason why I didn't fix the error with the end peices and that is that I'd actually already scrapped one version and started again. The thought of having to start again with another 4 days of intent concentration made me feel a bit sick.

 

Here's the first iteration and yes, you'll notice correct end peices. But I wasn't happy with the rest of the wood, between the two, I still prefer the rebuit texture.

 

WaterTowerMRdusk2.jpg

 

 

WaterTowerDusk4crop.jpg

 

 

The student version of 3ds max, nearly made me sign up to be student again, but as I work commercialy as a freelance, I would have had nightmares. :)

 

Yes, the lighting is calculated by the rendering engine. It's called Global Illumination and Final Gather. I just place a sun in a position to get nice shadows and variations in light, then the software shoots photons out from the sun and you tell it how many bounces you want it to have before it dies. It makes for nice soft shadows and hard shadows in all the right places.

 

It's also worth pointing out that the rendering of the picture, i.e. colouring it in and adding light can take upwards of several hours to complete. I did a job two years ago that had to be rendered so large that it took each image 17 hours to render and there was 24 images. I had 5 computers, each with multicore/multithreads running, totalling 30 processor threads and it still took that long. Had I had to render it on a single machine, I think it would still be doing it. :D

 

This image was probably a couple of hours rendertime, not very long in comparison.

 

Anyhoo, hope that hasn't bored too many people to death. It's not very often I've had the chance to a talk about a subject that know a little about.

Link to post
Share on other sites

  • RMweb Premium

Not at all. Fascinating thread!

 

It's like professional audio recording software - I could make anyone sound good!

 

 

Best, Pete.

 

I think to make my voice sound good you would have to destroy the file and start again!

Link to post
Share on other sites

Hihi, I also use Cubase 5 and Pro Tools. I'm a qualified audio recording engineer too.

 

Autotune anyone?

 

I'm a Logic Pro and Garageband guy. With all the add ons (fancy a Fairchild Limiter program?) you can easily exceed $5K.

 

I'm not a qualified engineer although I have worked at and with Abbey Road - I now mainly classify myself as a musician - which is where I started way back with Decca Records UK in 1967 when I was 16. EMI/Capitol Records sidetracked me.

 

I don't know any other life than being in the Pro-Music Business, for good or bad.

 

Best, Pete.

Link to post
Share on other sites

Archived

This topic is now archived and is closed to further replies.

×
×
  • Create New...