Finished!!

Finally got it all through! today was a simple case of tweaking the editing slightly, coming up with a new name for our guy as not everyone agreed on Pony Fontana. This is fair considering a) he doesn't look cuban, and b) we didn't really want to just rip of an existing name. (Tony Montana.)
Anyway, it's all rendered out in quicktime. I wanted to post it up on the group blog to tie everything off, but blogger wont let me upload anything above 100 mb. Never mind though, I'll probably wack it up on facebook or youtube instead!

One last time!!

Foreground layer works perfectly! Shadow layer however, does not. I basically need to render it out again only set the porch roof to cast no shadow, as according to the current layer it's being cast on the ground. This means it looks wrong when layered on top of the mansion. We tried removing this on After Effects, but it would get far too complicated having to keyframe a layer mask. Not to worry though, I've sent it to the render farm to run overnight, so in the morning it should be ready to go. Plus, Dan's been working hard on compositing and editing, and we now have title sequences. Basically everythings in place and nearly ready to go minus some minor tweaking, so we will easily be able to hit the deadline tomorrow at five. I got to admit, I'm very excited!

aaargh more re-renders!!

Ok so while the door now nicely sits in the middle where it should be, the foreground statues seem insanely bright and the trees/statues cast no shadows on the ground. This is probably because the ground is on a seperate layer to the objects. So basically today I've had to send 2 more layers to the farm, one being a shadow layer, (had to introduce the ground into the layer, but am still only rendering the shadows so it shouldnt cause any compositing issues), the other being a new foreground layer just with the incandescence turned right off and the ambience turned down on the effected objects.

Hopefully these will come out today.

Lolz!!

so so nearly finished, I cant believe what an obvious mistake I made! I sat through the whole hour and a half it took to render the mental ray layer of my scene only to realise that because the porch was included in this layer the door floats in front of it when compositing. Haha! so today I couldnt really get much done other than sort out the layers so the mansion and hill are the background and the porch, statues and trees are the foreground. This should mean the door can sit nicely in the middle where it belongs. Unfortunately due to the vast que on the render farm I've had to leave it overnight! will see what happens tomorrow.

Epic day

Sorry havnt posted for a while, but heres the lowdown:

Yesterday me and Dan discovered there would be some issues with crossing between rendering softwares when the mansion exterior becomes the foyer. Getting the doors to match up was excrutiatingly difficult and we spent too long trying to get the textures and lights right. Today it became evident that the solution was far easier then we thought, and all we had to do was render the door and only the door out in maya software (with a couple of tweaks to the texture and a light set to only affect the door) so that it matched Dans. Everything else in the scene will still be rendered with the mental ray settings. These two layers can just be composited together afterwards.

The next issue that arose from this was that the door couldn't have a shadow on it as again this would make the crossover obvious seeing as Dans door has no shadow. We spent ages changing the lighting to try and get it so that the roof cast no shadow where the door was- we played with the actual light and even tried setting the roof to cast no shadow at all, though that looked weird. Again the solution is far easier than we were making out, as when it comes to compositing the door wouldn't actually be effected by that shadow anyway as it is on a seperate layer to that light, and the colours would simply appear to be what the colours look like under the influence of a shadow.

Next step was to slap down the new musical score down (with thanks John Solly) and adjust the camera movements slightly to fit the rythm better. No problems here.

Next I had to deal with the sky. This was easily enough created with a dome textured a pale blue colour set to recieve no shadow. This did mean I had to then adjust some of the textures on the mansion as they now rendered out darker than they should be. The fact that it was a dome meant that Phils image plane wouldn't work well on it, and it had to be a dome because the camera angles pretty much cover all of the sky. So I had to find another way around the clouds and hills. Photoshopping the image plane so that the cloud was a seperate image was the obvious first step, however applying this to seperate faces did not work. (The blue sky colour of the image is impossible to match up to the colour of the sky dome.) So instead Nate showed me how to cut the shape of the actual cloud out on photoshop and then tranfer it as an illustrator file. This enabled me to import the shape of the image as a curve into my scene, in turn enabling me to create an object to the exact shape of the cloud. Then it was a simple process to uv map the image to fit the object.

Once I had the clouds in the air they required some transformations to get into the compositions on the various major shots. This also required some keyframing to make sure they stayed looking good as the camera moved. For the shot where it holds for a few seconds, Dan suggested a litte bit of floaty animation for some cartoony cloud movement. This worked out well.

Finally I needed to deal with the hills, and this time the solution was far easier than doing complicated processes with photoshop or the image plane. All it required were some smaller spheres surrounding the main hill taylored to fit the film gate. The colours of which were identical to the colour of the main grass material, but getting paler with distance. And I gotta say it looks fantastic. Especially with the shadows cause by the mansion. (Although I did need to remove the shadow from one or two.)

Phew, I think that's about it. I have sent it off to the render farm and am currently waiting on it to come back. I should then hopefully be able to post the batch render perhaps with sound so you can see everything I'm talking about. Im seriously anticipating it, as this will be the first time I get to see how the transition between exterior/foyer works out, and if the compositing of the door works as well.

Worked closely with Dan on all this today as lack of computers ment we were sharing for most of the day, so thanks to him for helping me out so much.

Problem solved!

So the solution to this rendering problem was so blindingly simple I'm still kicking myself. Rob helpfully pointed out that if the rendering works through the perspective camera but no others, why not just animate the perspective camera? Im such a foool!! Of course I would have liked to have understood exactly why it didn't want to work through any new cameras, but with the time we have left I think it's best to just leave it be for now.

Here's a quick playblast of the updated camera movements. It means we get a better view of the statues out front now, which is good.

On a completely seperate note, Dan pointed out that I didn't have a folder with my textures in, and in fact the location of my bump map texture was set to the public folders. This was going to cause problems later on, so I had to save the texture into a newly created texture file and reload the location of the file in the scene. I also finally got round to deleting loads of old files, opening up so much more space for saving and generally keeping my project folders tidy.

Next steps are to get the image plane up for the background, maybe do some soft modification on the ground, and try and apply a bump map to the path. Other than that it's nearly render time!

Camera test

With help from Dan and Phil I finally got a half decent camera test playblasted. I hate maya so much though. So much. The introduction of a new camera into this scene has completely messed up my mental ray rendering. All the settings are the same, and the render time is the same, but the quality is Maya software standard. It makes no sense, but Rob's in again tomorrow so I'll get some help to try and sort this out.

I also finalised the second character rig today, with the jiggle deformer and correct shoulder weights. It felt good :-)

thong rig

I spent today on the rig for the thong model. Not too many problems here, though the model needed adjusting slightly so that the rig from the previous model could fit. Painting weights was not a problem this time. In fact I'm glad I had the opportunity to do it again, as it has completely cleared up any uncertainties I had before about what I was doing wrong. I have the workflow nailed and once the penny dropped I encountered no issues I couldn't solve.

The butt clench however proved to be a bit more of a problem than I first thought. Skin clusters can be difficult to control when it comes to parenting the control object to the root controller. Origonally I had two objects controlling the respective cheeks which were then in turn controlled by a master butt control. I ultimately had to ditch this, and resort to grouping the butt clusters together, and linking the scale x factor to a new attribute on the butt control. This provides the necissary butt clench, and the butt geometery itself doesn't deform when moving the root controller. However, this has also meant that the clench is entirely linear, and does not clench to account for any angles other than the standing pose. I don't think this should be a problem as this is probably only going to be used when hes standing straight anyway. I might try and work something out that might be a bit more diverse tomorrow, but until then here's a quick playblast of the control.

Door developement

Ok so having decided we would probably do the blendshapes when and where we need them, I've had time to continue work on the front door. Basically it was too plain for an object that's going to be seen close up. I had an experiment with the planks on a one in / one out pattern, but it looks bad, so instead I drew all the planks out and defined the gaps between them. This looks a lot better. Then there was the question of the wood texture. Mayas default 3d wood texture looks aweful no matter how much you play around with it, so I downloaded a better one from www.cgtextures.com. It is subtle and looks so much better.

Then I had to play with the gold texture a bit more, because in the current lighting for my scene it shows up a nasty brown colour. In playing with the abient colour and reflectivity, I inevitabley made it look brighter than the sun, but after further playing it looks ok.

I also took the reflectivity off of the tile material, so the porch roof isn't so shiny. I also had to use the automatic mapping function on the small steps, as the texture wasn't quite working.

Lastly, I switched the statues around so that they point towards the door rather than away from it. (It just makes more sense, and leads your eye into the door more.)

I'm pretty happy to call the door finished now.

Statues

Ok as I'm still pretty excited about the rig I decided to do one more thing with it before I call it a day. I really wanted to see how the statues look on the plynths at last. Basically I repositioned the rig into the statue position and duplicated the geometery, deleting any unnecissary parts of the rig to keep the scene clean. Then it was just a case of scaling it negatively to flip reverse the duplicate making for the nice mirror statue model.
I decided to use the same material that the pillars are textured in, because even though it's plain white the global illumination makes it look marbely. This is open to change if anyone has any good ideas, but one way or another I think the pillars and the statues are made from the same material. (Whatever that material be.)
I'm still inspired by the engraving Dan pulled off on his statue plaque, and I'm now wondering if these statues should have similar plaques made in stone. Looking at it the plynths look very very blank, and could perhaps use a plaque with "P.Fontana" engraved on it or something.
Comment any suggestions about this!
Or I could just talk to you both about it tomorrow. That's probably the best plan actually, considering I seriously doubt you'll be reading this before then.

Full complete rig

Ok I admit the title of this post is slightly misleading as there are no blendshapes yet, but I spent a lot of time today fine tuning. I sorted out the weights on the hands so they deform as they should. I also sorted out the afro with a jiggle deformer for automatic secondary animation, and also fixed the zip up with a controller. I also put a squash and stretch deformer on all of the geometery so Dan can give him some bounce when animating. This is controlled by the sphere with one attribute linked to the squash factor of the deformer. Min value -0.2, max value 0.2.

As for the problem of the face dissapearing when making the rig controls invisible, this was solved way easier than I thought last nite. It's a simple case of selecting the individual objects and turing off the visibility in the object display.

I had planned to get blendshapes done today, but made a fatal error by saving over the file I had been working on during the process, rather than making a seperate file to save the blendshape tests to. This mistake fully punished me when I found the rig had been de-skinned for some reason, and due to my frequent saves was unable to go back to when it was working. Luckily for me I had a seperate file from the beginning of the day, so all I had to do was repeat my whole days work to get it back to where it should be. This time I have 3 versions of the file saved, one to my memory stick, to ensure that this doesn't happen again.

So while I didn't quite get the rig fully ready to animate with today, it's still very nearly there, and I honestly think it's my best rig to date. I'm very pleased with myself on this one. Tomorrow I have to sacrifice the day so I can get my case report done and out the way, but over the weekend I plan to try and get the blendshapes done. I at least want to be able to give him to Dan and Phil by the end of monday, as all these rigging hiccups have set us back behind schedule slightly. Other than that, good days work!! I can sleep at last :-)

Yes another rigging problem

Ok so I was just trying to do a quick rig on the zip, connecting the rotations to an outside object. The good news is, this works perfectly. The bad news is that in fiddling around I have discovered a new problem. Having parented the non-bound eyes/mouth/zipper to the root joint, this in turn means it is parented to the pelvis control. This in turn means that if I want to make the rig control layer non-visible (which is essential, otherwise there will be objects floating in front of the character in the film), it also makes the facial geometery invisible.

I'm not going to try and solve this now, simply because I really want to go home.

Painted Weights

The main bulk of today was spent sorting out these problems with painting weights. Many tears were shed, and I nearly ended up owing the college money for a new computer, but with some moral encouragement from Dan and some very useful help from Alex Hulse I didn't have to resort to breaking anything!

The first issue that had been causing me problems was how I'd bound the skin to the character group rather than the individual geometery. This was probably confusing maya and causing the duplicate influences, but was easily solved simply by re-skinning.
The second issue was some keyframes on the mouth left over from Dans blendshape test. Again, easily solved by deleting those keyframes.
The third and most complex issue was my workflow. Alex suggested that I should focus on adding influence to joints that shouldn't be influenced by anything else so that nothing else can influence it. For example, completely flooding the afro geometery white so that no other joint can mess it up. After sorting out these main areas, it was just a case of fine tuning the other more delicate areas like the hands and feet. I still don't fully understand to be honest, but I definately have a much better idea of how to tackle painting weights now.

Another thing I picked up from Alex was not to bind anything that's going to be blendshaped or rigged further. I.e. the zip, as this will disable the ability to do anything with it. Instead I have parented them to the root controller (pelvis) so that they will still always stay with the model.

So here he is in all his arrogance! I'm very pleased with this rig. There's still 2 more things I want to do. One is to put a jiggle deformer on the Afro geometery, and the other is to sort the zip out so it can move. Both should be fairly straightforward processes and shouldn't take too long.

I havn't shown everything the rig is capable of in this playblast, mainly because it seems like a waste of time to go about keyframing all of that. Basically the full description of how it works is a couple of posts down.

Change of scene

Gotta give credit to Dan for this idea. Basically rather than having the mansion set in an open plane with hills surrounding, I've changed it so it perks right on top of a stupidly steep hill. This not only fits in more with the origonal idea of having the mansion atop a hill, but also eliminates growing concerns of how the image plane would work. With the last ground plane, you would still be able to tell it's just a flat plane even with an image plane in the background. With this one, all I will need is a plane showing more hills surrounding this one. This should ease any restrictions on camera movement I might have had. It also looks more cartoony and fun, and has reduced my poly count immensely, speeding render time to 2 minuits 30. Great idea Dan!

I will still work into the ground closer to the mansion with the soft-mod tool as before, just to add some depth to the closeup shots.

Rig notes

here's those notes I promised on blendshapes, painting weights, the zip and the fro. Also slightly off topic are the quick sketches I made to help visualize the remake of the door.

Cleaning up the scene

I decided it was time to reduce poly-count of the scene, as previously it was just not acceptable. My biggest mistake on this whole project was deleting history after smoothing and saving, making it impossible to carry on working with a light scene. I did this without thinking and all I can do is learn for future projects. As you can see though, I have taken away every face and tile that is not needed in the most revealing shot of the mansion. This has reduced my mental ray render time from 8 minuits 40 to 5 minuits 30. Big improvement. Rendering out in layers should also help reduce render time when it comes to the crunch.

I also breifly added some soft modifications closer to the door before the tool stopped working for me. I'll probably ask alex about this tomorrow as well.

Door frame texture

Today was one of those days where I worked loads and got very little done. Using Hyper Realistic Creature Creation, (Alias), I attempted dealing with the painting weights using various reccomended MEL scripts and tools such as hold weights. All of which resulted in failure. However, I managed to secure a meeting with Alex tomorrow and he assures me he will be able to help.

So once again I ended up working on the door. Tried out the textures I wanted to and suprisingly I think the gold texture works a lot better than the tile texture. I decided not to bother experimenting with the wacky planks, as it's not really necissary and I'd rather save myself the time. I also pulled out the steps a bit more. I think it looks better, but it's never to late to change it.

Door in the scene

Here's the new door in the scene. Looks pretty good with those colours to be honest. I still want to try applying the gold texture to the frame like Dan did with it in his scene, but I'm predicting it will stick out like a sore thumb. In case of that I'l try the roof tile colour, as this might link it up nicely. (The roof tiles, particularly on the porch look quite golden. Might be able to get away with it.) Otherwise The colour of the frame is quite nicely fitting with the mansion exterior already.

Also I really want to try stretching the tiny doorsteps out a bit more! They are far too flat compaired to the other steps. It looks weird

Open Sesame!!

Here we go, quick change of pivots and removal of non-keyable attributes, and we have a door that works!!

Front door remake

Ok, so today I spent at least an hour with Dan Dalli trying to work out exactly why painting weights refuses to work. All to no avail. So I ended up sending an email to Alex Hulse in the hope that he might be able to help out. (I will still post up some diagramatic notes For the weights and blendshapes in a minuit So you can see what the plan is.)

In the meantime I decided to get to work on the front door, as it's going to be one of the first things you see and at the moment it is very unimpressive.

The bottom one is the origonal remake- it looks far better than the origonal, but was still not quite right. Dan suggested I make it wider, and put the door knockers right down at the bottom to account for Pony's pathetic size. This looks so much better, as you can see in the middle image.
The top was an experiment to see if pulling planks out in a wacky fashion looked good. As you can see it really really does not. Far too busy, so the middle one is the model I will use

Bouncy

Been working pretty hard on the rig this weekend. It's not finished yet, but I just had to post this test playblast simply because it's one of the funniest things I've ever seen. I almost don't want to paint weights!! (I will of course.)

Basically what I still need to do is sort out the orientation of the wrist joints, as at the moment they do not correspond to the rotation of the hand controllers. You can still rotate the hands, but its not very user freindly.

http://www.johnson-mortimer.co.uk/SJT_Full_Tutorial_List/SetupTutorialPages/Steven%20J_%20Tubbrit%20-%20Character%20Rigging%20Page%20005.htm

I will be following this link for the continuation of the tutorial, mainly so I can understand how to get a low rez version of the model to animate with that can be turned into the high rez model when it comes to render time.

Other than that it's pretty much a finished rig. (Painting weights and full blend shapes aside.) I decided not to give him a flexible spine, as he is pretty much a walking head, and a bending spine within that would look weird. I also decided there was no point in placing locators for the eyes as he has no pupils.

I've locked off every attribute that will be useless to make animating and keeping the graph editor tidy easier. I also created new attributes to control the fingers/thumb/feet/elbow movements. Im pretty pleased with it so far, as there should be no need to ever go into the skeleton during animating.

Rigging tutorial

Tutorialized is an excellent resource for easy to follow tutorials in any programme, including maya. Here's a link to the tutorial thats helping me with the character rig.

http://www.tutorialized.com/view/tutorial/Character-Rigging-in-Maya-Part-1/24521

Camera movement

Heres a quick playblast of a rough camera movement. Obviously needs a lot of work, i.e. timing photography. From this playblast I can see I might need to work something out about the edge of the plane where you can tell the scenery suddenly ends. Background image planes will be useful so I can see exactly what I need to do to the actual model if anything.

Grass test

Ok, this has been way more stress than it's worth. I followed that tutorial and fitting grass into the scene is basically not going to happen. The plane is far too big, and fitting grass with a suitable density into it causes all sorts of problems. I.e. I would have to smooth off the whole plane creating unecissary polygons, I would have to have it on an entirely seperate render layer, and even then it would still cause the computer/render farm to crash. A small patch such as the above is do-able, though it can't really be rendered with mental ray, and placing patches all over it is quite a lot of unnecissary work.

So I've gone for the simple solution of ditching the whole fur/grass thing. The realistic grass it creates doesn't really fit the style anyway. Instead, Dan suggested I go to work in finer detail with the soft modification. I agree.

focal length and angle of view

With some simple experimentation in the perspective camera, adjusting the focal length and the angle of view, I am able to come up with some interesting photography.

grass texture

Quick link to a tutorial I plan to try, using fur to create grass.

http://cg-nation.com/articles/maya-grass-tutorial-needs-maya-fur/

mental ray tests

I used Phil and the tutorial posted previously to do some mental ray tests, specifically focusing on global illumination. Under the current settings it takes a looong time even just to test render, but the results are immaculate. check this out compared to previous renders. I also wacked a grainy texture onto the tiles and walls. The top image, being the latter, has a colour adjustment on the grass as I've had endless comments about the grass on the others being way too deep.

global illumination tutorial

http://tuuhia.freehostia.com/gi_tutorial.htm

heres the link to that mental ray tutorial guys.

full colour test

Soft modification and rearrainged trees

In response to Dans suggestions I re-arrainged the trees to be closer to the front of the building. I also applied his colour suggestions and I think it works quite nicely. Might still need some fine tuning and definately a grainy texture. I also used the soft modification tool to create some hills for the land surrounding the mansion. next step is to colour the ground.

early colour tests

so here are some early colour tests. No textures thus far, just trying to get the colour right. The first image is totally wrong, its screamingly loud and all I can see is orange tiles. The second image is simply way too purple, and is too rich in colour. The third is getting there, adding more red based the colours from Phils research into the villas. The fourth is more like what I was aiming for, adding a bit more orange to get that salmon colouring.

The balconies and gutters are still completely the wrong colour, they stick out way too much.

Also, I perhaps need to seperate the actual glass planes of the windows from the mesh and do something more 'windowlike' with them. (I.e. transparency? lights coming Through? Simple reflectivity? At least a different colouring.) Particularly with the windows on the side sections as they are completely lost in the colour having no window frames or anything.

updated window frames

so we all agreed that the windows looked too "churchy" and maybe it would help to put a frame round the outside of the crossbars. It definately made a difference, and now looks much better..

window bars

quick update here. some added window bars. They appear a little bright here. will sort that out soon enough though.

Old stuff

Its taken me long enough, but I finally got around to scanning some old images in. The quality is aweful, but its just possible to make out some of the early thumbnail sketches for the storyboard.

The chair was drawn from the Victorian house on the drawing trip. The item interested me, because it's swirly design fit our style guide perfectly. Since then we have of course decided to include no chairs, so never mind! here it is anyway.

Also up there is a sketch I made of Pony Fontana after Dan had already designed him, simply to give me a feel for his proportions.

Balconies

Yup went for the balconies and it looks good. Had a discussion about the window shapes as well; the thinking was that the bottom two windows needed to be the same shape as the top three, but I think I'll leave them as they are as it ties the whole of the bottom floor together. As a compromise though, I'll probably put balconies on the bottom two windows to link them to the top floor.

Windows/Gutters

So no problem with the booleans difference at college. Must be a problem exclusive to my laptop. The indented windows have meant that I now have to work on a smooth mesh, as trying to smooth the mesh of after using the booleans difference led to countless issues. It shouldn't matter anymore though, since I have no reason to work into the building any further.

I Added some guttering under the roof as well.

I'm not exactly sure right now whether or not to add window frames/balconies, or even how to go about doing it if I do. Will experiment and see what I can work out.

Pathway with trees

Ok so what do you guys think about this? On the plus side, the uniformed trees lead the eye into the mansion making for very nice photography. On the other hand, there is no boulivard in the world with trees that all look identical.

The path and ground is all one mesh, but I checked the uv map and its very obvious where the path is. Won't cause any texturing grief. I will prob use the soft modification tool on the grassy bit as well, make it look more natural.

Rounder edges

From the last render I decided I needed to make the edges more obviously rounded to fit the style guide a bit more. I also deleted the wall with the door from the porch model as it obviously stands out on the model. Instead I will create the door seperately. Here it is. I think it looks better.

Roof tiles

Heres the mansion exterior with the roof tiling. Had to do this the long way as I couldn't get the duplicate special to work well enough. The roofs on the side sections for example are sloped in an unusual way, requiring manual tweaking. I did have to adjust the roof shape slightly to get it to work. Duplicating entire lines of tiles however still sped the process up.

Dan, I know you reccomended texturing the tile first, but I grouped all the tiles together, so I can still apply one texture to all of them at once.

Re-re-modelling the mansion

After getting to grips with this simpler technique for modelling smoothly, I decided it was in everyones best interest for me to re-do the mansion. On previous models I had taken an organic approach to modelling as I have previously only modelled characters. This approach led to the model looking rough, and gave unpredictable results when smoothing the mesh. The new approach is very systemactic, and is far more controlled.

Already this mansion is looking much more pleasing to the eye than the last attempt.