Monday 18 June 2007

hefty update from alex

Hey andrew,

Apologies for not updating sooner with any models but i have been spending quite a bit of time experimenting with texturing, displacements and modelling techniques over the past few weeks but i will fill you in on all that as i go through.
First things first, i cant remember if i put my final concept art up on a previous post so here it is...

Im going to consider the background a bit more later on, but essentially its going to be sky without a horizon and im going to experiment with colour when it comes to lighting/rendering/comping stage.

Anyway, i did a proof of concept scene to test out the textures/displacements etc. which you can see here http://www.youtube.com/watch?v=DOtApw63fow if the video below doesnt work.





















You cant actually see any of the textures very well in the wonderful youtube quality but it was an experiment really to see how the displacements/textures reacted with a moving camera. To streamline my workflow I wanted to ascertain how much i could get away with using low quality and when displacements actually made a difference. In production i guess you generally use displacements like textures and scale them dependant on the distance from camera?

Here are a few images for you to peruse




Im ok with how it rendered out but i used the same (fairly low quality and rough) texture on all of the branches but they are far from what i would like in the final product. I want the stone structures to be quite battered and worn and i think it was working pretty well with the arched stairway. The design of the windmill blades is kind of work in progress, they are a bit sharp and straight edged but im going to roughen them up a bit.


I have been experimenting with quite a few ways of modelling the branches and what would be most efficient in my workflow. Initially i was starting off with a box and then just extruding faces, slowly building them up around the buildings. The image below was of one of my earlier attempts at roughing out the scene




i found that it was quite slow modelling the trees in this fashion and also, as each model became quite complicated, setting up the uv maps on each unique branch was going to be a nightmare.

The image below was my solution, having each branch separate, setting up the uvs first and using a bone chain to deform it into position.






Setting up the scene was about a hundred times quicker but the only problem i found with this is the joins between each branch object. In the test scene that i rendered it didn't stick out to much so i think i might be able to get away with it.


Onto the buildings now....


the main things im really stumped on regarding architecture is whether i should be trying to keep certain parts separate or if i should joining them together to form 1 mesh. For example if on the side of a building i have a window sill or frame, should i make that a separate object to the building? If i merge the objects together then i create an n-sided polygon on the wall which gets messy when i subdivide. not sure if thats clear but i am sending you a few objs which might explain a bit better. building_01_no_jitter.obj is the building i used in the simple scene above (without the deformation to its shape) where i built the window frames into the same mesh but tried to maintain 4 sided polygons, building 02.obj is where each part is separate.

Also the bell_tower.obj is another example where i have many separate objects for one building, should i be trying to merge these into one mesh and if so, how would i sort out the topology so that it wasnt a complete mess?


Anyway, another screen shots for you...


This is the overall scene that i am slowly building up, i know you cant really see too much of whats going on from this screen shot but its still wip. The plan is at the moment to get most of the buildings and branches roughly in the right place, then take a few sections and model/texture them in higher detail for some close ups. I did a block test which i would show you but it has been scrapped basically because there was too much camera movement (im now keeping things much simpler) and it was based on the layout of an earlier test model of the scene. Once i have done the new version of the block test i will post it for you to check out the whole scene in greater detail.

I think i have waffled on enough so i shall bid you adieu. Again, apologies for not putting anything up sooner but my modelling proper is going ahead full steam now that i have finished experimenting so i will send you more as soon as poss.

many thanks, alex

No comments: