Well, I’m gathering this is it. This is the end of my uni course and after this I’ll no longer be a student which is weird to say. This year has some what been a learning curve and I guess there are still things I need to try and resolve. For one, my lack of confidence gets the better of me. When I feel I can’t do something, I seem to make it final in my head that I’m no good at it and I never will. That alone stops my progress in my work and general self development. There’s been times where I’ve been consumed by anxiety, but I’ve had to put that aside and learn to just get on with it. I need to have more faith in myself. Once I managed to focus, I feel I managed to get my head down and work my way through. Through the past three years I haven’t found a specialism and instead I’ve always worked on the jobs that other people don’t do. In the end I became a generalist even if it wasn’t by choice. This year I thought I wanted to become a matte artist but I’m not so sure anymore. I think what put me off is the constant use of photos but I feel I need to experiment with actually painting over the photos. Again, it’s because of lack of experience that I’ve not narrowed down on a specialism but also why I lack confidence. That’s something I need to build up by honing my skills and deciding what I want to do and what I need to do to get to that point.
One benefit of being a generalist is that I’m glad I’ve managed to fill in my folders. I may not be strong at animating, but I at least managed to fill my artist and technical folders. This year has made me learn more about rigging and I’ve also learnt to composite for the first time, particularly in Nuke. As I’ve mentioned before, it has also been my first time doing matte paintings and projecting. I’ve learnt a lot of skills this year that I can bring with me out of university. I think this has helped me to understand the animation pipeline a lot better. When speaking to Amir for IPP, he suggested I look at a producer/director role due to my experiences. That’s something I can consider since I am quite keen on being organised and I have been a temporary director when Deon’s needed a break. Another area I want to look into is concept art and I have already spoken to a few so I know that concept alone will not get me a job. Again, my experiences should help me to maybe offer more in these particular job roles.
This film has not been an easy project, particularly as there have been family problems within the group (much like there was last year) and there have been times where there’s been a lot of conflict where people have had to walk away. The thing is, everyone comes back and continues working even if it’s not been easy. Although we’re friends, we’re also colleagues so we have to accept our responsibilities. I’m not sure if I can say I’ve enjoyed every part of it, it certainly has been difficult but we have had our fun moments and managed to laugh even during tough times. I think with this film, we’ve managed to show we’re capable and although there are mistakes that we wish we could correct, I still think the film is quite impressive. We need to remember, we did this as students and we’re still learning. I’m hoping to set myself a small project between now and degree show where I improve my concept art since it’s the quickest and most accessible right now. After wards I want to try different areas to see what I enjoy most and as I mentioned, to hone my skills. I hope you enjoy our film, we’ve put our blood and sweat into this! (not literally but we did sacrifice sleep and laptops).
Since everyone was busy with their roles, I decided to help out by applying textures to the models. Me, Ash and Niko all took turns applying it to the buildings in the background and afterwards I had the task of doing Tower Bridge, London Eye, and then the cogs for both Big Ben and London Eye. I simply had to assign a v ray material and add the diffuse, spec and bump correctly. I did initially assign from the wrong folders for the Tower Bridge unfortunately so when the file was transferred Tower Bridge lost its textures. However, I did learn from this and did the rest correctly.
Deon had some missing textures, so this is why some sections have been left untextured.
Deon requested that one of the things I look into would be depth of field although Gerome decided to also look into this in After Effects. I managed to find a nice tutorial ob Vimeo: http://vimeo.com/29414230
It was a lot simpler than I had imagined as it only required 3 nodes, shuffle, shuffle copy and Zblur. It was important that I had rendered in passes in order to have a separate Z Depth pass.
This allowed me to change the focus between the foreground and background although one thing I need to look into is be able to choose where I want to change the depth of field. I learnt from this tutorial that the Z blur node should always come last since if it comes before, the foreground image will bleed over the edge, at least the shuffles prevents that from happening.
Update: Unfortunately it has been decided that Nuke will not be used for the film. Due to the fact it is a software that is new to us, Deon thought it was better suit if we used something we already know like After Effects. Since Gerome and Deon having been compositing with After Effects for years, they’ll be the one compositing the film with After Effects.
So from the beginning we knew we would be using our laptops to render our animation. Particularly since the render farm is so temperamental. However, Deon decided we should combine all our laptops to create our own render farm. In total we had 8 laptops and 2 pcs were linked together through a switchboard. Deon’s pc was the main one as the other laptops and pc lent their processing power to render. It hasn’t been working perfectly, in fact far from it. It’s something new that Deon had tested out and we have had a lot of trouble with it. I hope it’ll run smoother.
My animating role was to animate the robot I rigged. I would have used Niko’s one but there were some problems. Well, for me anyway. No matter what version of Niko’s rig was sent to me, one of the controls would NOT work so I used my own rig. I guess this also demonstrates how my rig works. Initially I was done and dusted with my walk cycle, but then Ash and Niko had problems cycling the walk. Ash told me that I need to keep it to exactly 25 frames for it to loop perfectly.
First walk cycle
Due to the problem of being unable to loop my animation, I had to redo my walk cycle unfortunately. Ash and Niko gave me tips as well as video recommendations.
Niko recommended me this video which made since it is from the Animator’s Survival Kit
In all fairness I knew about the logic of back legs moving in the opposite direction of the front legs from term 2 when discussing with Lily about animating a spider.
Ash also gave me some videos from Digital Tutors where one particular image was very helpful
This would help me with the front legs and I guess all I had to do was do the opposite for the back legs. However, I found the back legs more complicated than I thought.
If I had more time I would definitely have sat down and learnt how to animate properly. It’s something I’d like to learn because I always think about how living things move and the secondary animation. I am a bit disappointed, but I guess I need to improve time management or my group on a whole needs to. As soon as I was given the model and finished rigging I did try to animate straight away but I guess I didn’t allow myself enough time. If it was a normal two-legged cycle, this would have been fine. At least I’ve learnt from this for next time.
Another task that was given to me was to design the main character at the end.
When describing the character Gerome gave me three works, pollution, devil and works. The obvious choice in my mind for pollution was a gas mask so I tried to find some references for that. Since our film is called ‘Devil in the Works’ when considering the word works, I initially thought of builders. I had the idea of trying to incorporate the uniform into the design. In contrast, I also thought of businessmen. The reason why I considered these things are because in our city, in order to survive you must have the right uniform (i.e. the gas mask for the pollution) as well as having some sort of strength or intellect. Initially when designing this character, I imagined him to be quick and cunning.
As you can see, this character is slender and I could imagine him being agile. I have given him a body suit since his skin can’t come into contact with the air. On the back I intended to add the ‘Devil in the Works’ logo as if he belongs to a company much like builders. I imagined that some of the humans that did inhabit this city would belong to this organisation and be considered as the ‘devils’ for a higher power. This is how the humans would generally be dressed like in order to continue inhabiting. However when showing Gerome and Deon, it seemed that had imagined the character differently. They had imagined that the character would be bulky and hench, possibly even mutated from the pollution. Due to this, I decided to scrap this concept.
The previous one I was more looking into creating a realistic human being, but I realised that was not what was wanted. When designing this character, my initial inspiration was Mr Incredible.
He has a top heavy torso and yet humorously tiny little legs. I decided to do something similar with my character and that it would be more far fetched rather a realistic design of a human being. I still kept the idea of the uniform but instead of making him sleek and well dressed, he was ragged almost like a barbarian. I altered the gas mask a bit more and added an oxygen tank as well as deciding to place our group’s logo on the back of it. Deon suggested that I added mechanical limbs so I made the metal arm slightly larger than his normal arm. To elaborate on him being somewhat like an over grown builder, I added a toolkit onto his belt. To further emphasise on the ‘Devil in the Works’ theme, I decided to alter the gas mask with devil horns and more menacing eyes. For the final touches I gave him a few dirt marks or specs of blood. This concept design couldn’t be further from the previous one but I do quite like it. More importantly, the others loved this design.
I did an orthographic concept for Gerome (which I did plan to go over darker). However, he forgot about this and modelled straight from the the concept art so the model didn’t really end up very accurate.
I did manage to rig the model similarly to the robot models, but Niko was extremely displeased with the model and approached Gerome. Gerome admitted he didn’t feel confident enough to do a human form in such little amount of time so he remodelled the character to be a complete robot.
Niko rigged this character instead and it does look a lot better than his previous model. I am a bit sad that because of timing and lack of experience, my concept wasn’t fully realised. For example the idea of the uniform, the general overgrown bulkiness of the character but sacrifices have to be made under pressure.
Apparently Dan had spoken to the group before I had came in for the lesson as I was told that we should add some sort of characters into our film to provide us something to animate for our folders. This was quite short notice considering these things should have been sorted by the second term, but these models and rigs didn’t have to be sophisticated due to the nature of our film. There is more emphasis on the environment rather than the characters. Gerome had decided we would just add a few small robots, and the main character at the end that we had previously talked about adding for some sort of narrative at the end. Deon added that there should also be a giant robot in the background but it will not be in focus as he’ll change the depth of field. With this new addition to our film, I knew I would most likely be the one doing quite a lot of the rigging, although Niko is quite capable of doing it also. Due to this I had to stop learning Nuke and start rigging straight away since this seemed like the most immediate task to do.
The only thing I’ve rigged before this previously were a spider and spider legs for our compositing project and although the robot’s legs are somewhat similar to a spider’s I’ve never really rigged the more human parts. I figured I would take the approach of making the skeleton and merely adding IK handles for the shoulder to elbow to wrist, and the same for the finger joints. However, this was most definitely not working out. The IK handles connected to the elbow joints kept bending backwards which wasn’t right.
Hey, sorry to ask so suddenly, but do you know about rigging arms? I just need a basic rig for a robot. I haven’t rigged in like a year, and I’ve only had one day to rig these robots D: I’ve rigged everything but the arms aren’t working out. I’ve used IK handles though but I figured for something so basic it would be okay.
haha don’t worry send me with rig, cuz I don’t have time to do the full rig cuz I am leaving for FMX tomorrow so I am not sure if I will be able to do a long hours job today! But I shall have a look .
I dont get it, cuz if its a robot then you dont unnecessary need ik for hands , plus the mesh of the robot is 1piece. so I guess your attempt was not wrong I would put an IK as well if its 1 mesh. Normally robot rig mesh needs to be separate like palm, elbow, shoulder screw nuts ect…
I’m not sure how it was planned to look like either tbh
I literally just got the mesh yesterday and was told to rig it
but I was just planned to put a basic rig since they’re not vital characters
we just needed some sort of animation in our film so Gerome created robots.
They don’t actually have an important impact on the film.
I think, it’s okay for a basic rig since the robots will intentionally be rigid. I just don’t want the arms to go backwards I guess lol. I just want it to move somewhat normal.
Sorry for the long explanation by the way. I’m not quite sure how it was intended to look when animated. Even Deon isn’t sure.
Seen as this is a mechanical character, you may have been better off modelling it in separate pieces and using curves with Orient constraints to the parts that move. My suggestion would be that you remodel the arms and the parts that you want to move as a separate mesh and constrain it with controllers, which would be the easiest way. This looks very complicated for skeletal rigging.
yea I don’t mean to say you made a mistake haha but I think this model is hard to be rigged mechanically
sorry I cant help
Taking on board what Arpit said, I deleted the IK handles and decided to parent the joint to the control with an orient constraint. Instead of moving the controls, you would rotate. This worked quite well particularly for the fingers when curling them into the palm. However, the next problem I encountered was trying to keep the controls in uniform and to stay with the joints and mesh.
You can see that when I moved the arm, the controls would go flying off. I’m not exactly sure what happened with the last image…the rig may have broken although it was somewhat humorous in an annoying situation. I couldn’t contact Arpit anymore because he was in Stuttgart but thankfully James Waters was kind enough to help me through this.
Here are a few emails we exchanged:
Hi Dom,
I have attached a file which demonstrates how each curve is parented to the next. The knuckle controls of the fingers are all parented to one another, going down to the knuckles themsleves, which are in turn parented to the wrist. The wrist is parented to the elbow and elbow to the shoulders. In addition to that, make sure that the history of all the curves are deleted, the transformation is frozen and pivots are centred with the curves before you make any constraints.
I would also recommend creating drivers for the fingers as it is less laborious than animating each knuckle control individually.
I hope this is useful and let me know if you need any more help. Particularly if you want to set up an IK/FK switch.
To James
Sorry James, but the file got corrupted Also when you said the pivots centred, where should the pivots be? Do you mean pivots centred of each control, or the whole mesh? I’m not sure what you meant. Would you suggest I go back to the beginning to make sure I sort out these pivots first since I’ve already added orient constraint.
Ah ok, basically, what I meant was that when you select a control, the pivot should show up in the centre of that control. When I initially opened the file, I found that the pivots for many were very out of alignment with their respective controls. Yes definitley go back to the beginning and make sure each pivot is centred within each control. Hope that makes sense.
Here is the same file, this time saved as .ma, so you should be able to open it.
To James
Sorry about this but I’m having another problem 😦 I tried parenting like you said and it duplicated the controls I did delete the history and freeze transformations
Hey, that is strange, I have never come across that before. Do the duplicate controls have any impact on the joints themselves?
To James
Yeah it does, it seems to restrict the joints, as if it’s torn between the two duplicates.
Ok, Dom. I see you sent me two files; the more recent one, 026 looks perfectly fine to me, all the relevant constraints have been applied and each of the controls are parented so nice going with that. It needs to be weighted, so it will be believable as a mechanical character, but I’m sure you know that anyway. Yes, always use the default settings when you’re constraining; the only change you should make is checking the “Maintain Offset” box. Anyway, I’m glad this was of use; I’ve been working on here all day so I’m going to finish up for the night.
Speak to you soon.
After exchanging many emails back and forth and speaking over Facebook, I finally managed to sort out my problem. I simply had to parent all the handles together. The legs were already done as the IK handles worked perfectly fine. Before James I had asked Niko about the rig the day before but he was too busy. He decided to also rig the same robot and gave it to Ash.
However, when inspecting Niko’s rig, I noticed the control on the robot’s waist wasn’t working the same way mine had. When I moved the control down it would bend the legs as if adding weight. When I tried Niko’s, it dragged the whole robot down. I looked at the file on both Niko and Ash’s machine and it worked fine. Niko sent me a fresh copy of the scene and even Ash sent me his animated robot and it STILL would not work on my laptop. I wondered if my Maya be buggy since while I was working on these rigs, they would suddenly break without being able to undo. Thankfully I had common sense to constantly save and particularly as a new file (I had about 40 scenes of different stages of me rigging). I think the arms on Niko’s robot bent better and more mechanically as mine seemed to squish like a human’s arm. Niko didn’t rig the fingers though which is the difference between ours. I decided to make the most of that difference by making sure to position the fingers in a pose.
Here is a video of the rig I did for the worker robot
Although I hadn’t actually rigged the legs yet in this rendered image, you can see that I have rigged the arms and fingers which I think works.
After this one, the next robot didn’t seem as difficult due to the fact that I’ve overcame the many problems this seemed some what easier for me.
However, due to the way the joints were, I wasn’t able to parent controls to the hips and be able to create weight on the legs/ball. That was one thing I had much difficulty with. I also learnt from Niko how to take away weights when painting weights, this was something I’ve never really been able to do (I usually use the component editor but this time I used the painting weights tool since the meshes should NOT be influenced by other joints since they’re metal, not human flesh). That was the only regret I had about this rig and if I had time I would have tried to get around that.
Here are just notes I had taken down or copied from the book. I don’t claim this as my own information, I merely decided to copy it down so that I can absorb the information better.
Nuke is a 32-bit float linear colour compositing application.
Most compositing and image-manipulation programs are 8-bit.
8 bit is good for displaying colour, but not good enough for some calculations or operations
It may result in banding – inaccurate display of gradients, change in colour happens abruptly rather than smoothly.
32 bit allows for what is brighter than white (super white) or darker than black (sub-blacks), 8 bit colours would get clipped. E.g. Paper and light bulb in a room. Different shades of white = float. if they were the same shade = Non Float.
Linear means linear colour space.
Computer monitor doesn’t show an image as the image appears in reality <- I learnt this in the lighting masterclass.
Has a mathematic curve called gamma that is use to display images. Most often they have a gamma curve called SRGB.
Because it’s not showing the image as it appears in reality the image needs to be corrected.
Process represents basic colour management.
If a gamma curve has been applied to make a middle grey in an image, it will react different to colour correction.
Correct way to manipulate imagery – in whatever way – is BEFORE the gamma curve has been applied to an image.
Correct way: Linear Image > Colour correct it > Composite it > Transform it > then apply a reverse gamma correction as well as negating the correction you just applied. Nuke works like this by default.
In Nuke 0 is still black but white is now 1 to make it simpler.
Dynamic range changes from image – Image usually has black, white and everything in between.
Dynamic ranges are split according to their brightness values:
Shadows or low lights – darkest colours
Midtones – neither dark or bright
Highlights – Brightest colours
In Nuke there is also:
Super whites
Sub blacks
Colour > Maths > Add – Add operation adds colour to everything uniformly. The whole image changes together.
Colour > Roll off Contrast > Contrast Value low lights get pushed down and highlights are pushed up.
Centre Value
Colours above the Centre Value will be considered bright and pushed up.
Colours below the Centre Value will be considered dark and pushed down.
Similar to Multiply Node – Virtually the same
Lift is virtually the same as Roll off contrast
Lift – Value looked at 1, the farther the values are away from 1, the bigger the effect.
Multiply operation multiplies the dynamic range by a value. Multiply operation is virtually the same as changing the exposure in a camera or increasing light.
I/O Graph – Input vs Output graph
x axis – colour coming in
y axis – colour going out.
Perfectly diagonal line represents no colour correction.
Expression 1 Script
Reformat – defines resolution of image. this case it’s 256 x 256 > Ramp (from draw tool box) generates ramp, have a black to white horizontal ramp from edge to edge > Backdrop – Highlights areas in the tree (from Other toolbox). Tells you where to add your colour correction nodes in the next step > Expression (in Colour > Math toolbox) lets the user write an expression with which to draw an image.
Can do with an expression node:
Simple colour operations (Adding, multiplying but wasteful)
Complex warps or redrawing of different kinds of images all together.
In this it draws values of a horizontal black to white ramp on – screen a white pixels in the corresponding weight in image.
Gamma change midtones without changing black or white.
Multiply – Affects highlights more than low lights.
Add – Changes everything all together.
Contrast pushes two parts of the dynamic range away from one another.
Centre value at 1 exact same as multiply curve
Centre moves the low lights and highlights point.
Centre value at 1 – White point locked while everything else change – opposite of multiply.
Soft clip smooths out the edges of the curve so the colours don’t suddenly turn to black or white.
Images need to be unpremultiplied then premultiplied again.
Curve tool Max Luma Pixel – finds the brightest and darkest pixels in the image.
Image – Constant – Temporarily replaced black background with middle grey background.
CONSIDER ADDING THE REST OF NOTES?
Colour Correct Node controls for:
Saturation
Contrast
Gamma
Gain (Multiply)
Offset (Add)
Range tab shows graph
Highlights = White
Midtones = Grey
Shadows = Black
Green/magenta mix of two ranges.
This is from the files the book provides as exercises
This was before the colour correction, you can clearly see that the car doesn’t match the background plate.
This is how I managed to achieve this:
Car’s matching blacks and whites
Ctrl and drag Curve tool 1 Minimum Luminance Pixel Animation > Grade 1’s Black point animation.
Ctrl + drag Maximum LP Animation > Grade 1’s White point animation.
Ctrl + drag Maximum Luminance Pixel Animation > Gain
This has matched the foreground’s shadows and highlights to those of the background.
Viewer – Y – Luminance = Greys
Edit gamma with the help of multiply.
Still need to change colour – look into red channel.
Have to change multiply + offset values for perfect results – distance from car from camera means black point will be higher and white point will be lower.
So far I have been following the book Nuke 101 which has been pretty handy as it supplies exercises on a CD. I’ve been following each chapter although I may skip what is irrelevant to the project.
To begin with I was trying to get myself familiar with the interface as all software can be initially daunting. So far I’ve found it quite simple due to the fact that Nuke is node based so everything is visual. Maik had briefed me a while ago about the software and how the advantage of it is the fact that he doesn’t need to be by a computer to plan out his node flow. If he has an idea he can draw the flow in a notebook and take that to Nuke to fully realise it. Furthermore, he said that if he takes a photograph with his SLR camera, he can put in the settings he used for the photo into Nuke, much like V-Ray.
The above print screen is evidence of me following Chapter 2 which teaches about Basic Compositing. Through this simple node flow I’ve managed to learn about each node:
Read are the images/video clips you’re importing into Nuke
Sometimes the images you load into Nuke are straight and will have to premultiply. For example, the doll influenced the background because it was a straight image and therefore had to be premultiplied.
Grade node is the colour correction which you can only do to a straight image, hence why it has been placed before the premult node.
The Transform node deals with the size, position, rotation, etc and must be added after the premult node.
Shuffle deals with the channels. In the print screen the mask was in the red channel but it had to work across all channels in order for it to be effective and the Shuffle node allowed me to do that.
Merge enables you to merge nodes and in this example that means combining the foreground and background.
Write node is basically rendering the scene.
All this was pretty simple so far.
This is the result from Chapter 3 where I learnt how to make a more complex node tree. This was actually quite difficult for me as I’ve not handled render passes in the past since we’ve always just rendered it straight. I’m also not very experienced with lighting and although I know some of the terminology, I still have trouble quite understanding it.
Another problem I encountered seemed to be the actual book’s fault. I spent the night and most of the next day trying to make something work but it just wouldn’t do. It was only when experimenting with other options and talking to Deon that I realised it wasn’t my own error, but that of the book’s.
As you can see at bullet point 8, it says ‘choose Avs Intensities’ although that alone is a typo since it should be AvG Intensities as illustrated in the image above. Remember, the image above clearly says Curvetool1 (Avg Intensities)
A few pages later it changed to CurveTool1 (Auto Crop). It was having problems getting the Auto Crop Animation attributes to turn blue or to change over each frame. It was suppose to be Auto Crop, not Avg Intensities.
I’m still trying to understand this chapter better by going through it again and reading it properly. I didn’t want to just follow it and not understand the reason for it.These are a few notes I’ve taken down in my notebook despite the fact it’s already in Nuke101 as I want the information to sink in:
Render passes are always premultiplied, so you have to unpremult all passes to colour correct. This explains why on each branch there is always an unpremult following a Shuffle node. For each Shuffle apply each respective render pass in the properties. Plus = Add which means you’re adding (name of render pass) to the composite.
Lgt is set as the background of passes.
Shuffle, Shufflecopy and copy reorganise channels into a single branch to ease image manipulation.
Shufflecopy takes two inputs.
Shadow is a subtractive pass so you set it to minus. You’re taking light away.
When we attended the Master classes that were held out Ravensbourne, we all seemed to agree that Matte paintings were the best way to resolve rendering time. The class made me extremely interested in considering Matte Painting as a career as it somewhat ties in with my desire to do concept. In a way they somehow flow into each other’s fields.
I bought the ‘Digital Handbook Painting Handbook’ and I found it to be extremely useful. It actually teaches you in depth how to do a matte painting and you don’t realise just how complex the process is from designing to the actual projecting. For now I decided to experiment with the projecting.
I had each section in layers, i.e. the castle, bridge and sky. I had the over all matte painting as a single image as well. For this particular projection to add depth I would make each section 3d since the camera would not move much and I could get away with this. As you can see I just created a simple cube for the hill.
I sculpted it to get the general shape of the hill. I then made two more cubes for the castle and bridge and edited it to fit the general shape of each component.
I then created a semi dome for the sky and projected the texture, making sure that the sky image emits the lighting.
This is with all the textures applied. As you can see the bridge is a bit off. Right now it just looks like a plain image but I did a small camera edit to show that I have actually gone through the process of projecting.
My projection test
I’m really pleased with the outcome of this projection test but obviously I need to learn how to create my own matte paintings. I really do think this adds depth.
I did some more projection tests with a sky I created and just random cubes since we had the idea of using a dome to cover the whole city and project the sky as a texture. However, as you can see, this looks pretty much like a cheese.
However, I managed to sort this out and experimented with using two cameras. One to project the sky, and one used for the actual animation. I was worried that the objects would block the sky projection but it was fine.
Initially, I had no idea how to approach Matte paintings. These were the first tutorial I tried to somewhat follow:
Due to the fact that this film is to be realistic, it was decided by me and Deon that it would be best if I used photographs rather than actually paint my own buildings. Due to this I would find photographs of buildings and distort it to make it into something different.
I removed the sky from the original images by either masking it off, lasso tool/polygonal lasso tool or magic wand tool and replaced it with a different sky image. Here I tried to change the general mood by adding an orange overlay layer. Something that I learnt from these tutorials were to use a rough brush such as charcoal and just erase parts of the bridge or buildings and to change the size of the brush for variation. Due to the roughness of the brush this gave the impression of ruined buildings.
I can’t say I like this matte painting since it doesn’t look realistic. The light from the sky does not match the buildings and in fact the difference stands out too much. Over all there wasn’t enough applied to this that changed the image drastically.
For another test I tried to change the lighting more drastically by playing with the contrast, brightness, etc. As you can see here, the mood is much darker and Westminster stands out much less in comparison to my first matte painting.
When I posted this up on Facebook, Deon gave me a few criticisms and even highlighted where I went wrong.
Deon Carlos Litchmore I don’t understand why you’re doing a painting of the big ben?!
Deon Carlos Litchmore Secondly I understand the whole ‘paint’ in matte painting but it doesn’t have to look like one. Your compilations need to look photo realistic so you need to work on your edges and your strokes. Better than before but still a lot to work on.
Dominique Koomoshan So I need to sort out the edges of the buildings, make the destructive part not stand out too much, edit the buildings so that the light source is correct?
Looking back at this now I can see what he meant, the lighting once again did not match the sky at all. Also, from this point I decided to try and make matte paintings that could be used for the scene rather than continue testing so much with images I can’t use.
With my next matte painting I decided to take a completely different approach and instead of focusing on particular buildings, I decided to actually do a collage of buildings from different images instead. I literally deleted the background of every image and pasted the buildings together. The sky image is a HDRI that Ash bought and Deon edited so that when it is used for the v ray sky dome, it’ll create a nice variation of colours when the camera is moving through the city.
I’ve been using Photoshop CS5 for the first time when doing these matte paintings as I previously used CS2 so this was a huge leap. When exploring the settings I found the auto-align layers. I tried looking this up for this particular matte painting (link: http://my.safaribooksonline.com/book/photo-and-graphic-manipulation/1592004873/matte-paintings/ch12) but it didn’t work quite as well as I had hoped. However, underneath that was the auto-blend layers. I tested blending all the layers and it created this nice effect where the distant buildings seem to disappear in the horizon. However, I noticed there’s a giant brown box that appears in the bottom right hand corner.
I decided to go back a step and tried to make the buildings all a similar colour so that this would no longer be a problem by playing with the saturation, brightness and contrast.
Here you can see the outcome is much better.
Deon told me to take the matte painting into Maya and test it out in the scene. When I did this I realised that the perspective was completely wrong for it to be placed behind Big Ben. Although this meant I had to do another matte painting, I guess this could still be used for a different section of the film.
At this point, we had decided to change from mental ray to v ray so my projection tests previously became some what relevant, particularly the sky projection as there was a different approach to it that Ash and Deon had figured out.
Inspired by both methods I used, I focusing on a single image and collaging other buildings into the matte painting. Instead of using the whole of the HDRI sky, I print screened how the sky would look behind the big ben and decided to base the lighting and colour scheme of the buildings from that.
I heavily relied on this Youtube channel’s videos:
Photoshop tutorials explaining how to make a ruined city
These videos have helped me very much when it came to techniques that I feel I would continue using these methods until I found something better. I used the same erasing technique I used earlier but this time I added images of ruined buildings to fill in the gaps to make it more convincing. This can particularly be seen from where I’ve made a hole in one of the buildings. Moreover, I have learnt to use the burn tool to give the impression of the buildings aging. I had some trouble softening the outline of the buildings as I either blurred or erased too much. Deon taught me how to think about light direction and by creating a temporary sun to get an idea. As you can see from some of the buildings I added a white overlay for subtle light.
Here I have once again taken the matte painting into Maya to see how it looks behind Big Ben.
Deon had asked me to try and add lights to the buildings but we soon realised this did not work as the matte paintings would be influenced by the sky’s light and therefore sometimes being affected by shadows. As you can see here, despite the fact I used white, it is very much dulled out.
From all the techniques I have gathered while doing these matte paintings, I’ve somewhat mixed and matched methods to suit what I feel is best. This matte painting I just edited the saturation, contrast and brightness of the buildings so they’re more uniformed. I also added an orange overlay to suit the sunset of the sky. However, when I posted this up on our Facebook group, Deon said that the buildings were far too uniformed and not random enough. Moreover, he said the buildings looked too similar. They lost their individuality.
As you can see here, I made the buildings a lot more random as if they’re sinking in on one side. I do feel this makes it look more like ruins that have been abandoned. I also decided to add rubble and telephone pole images to add to the desolate atmosphere. This time I didn’t edit the buildings to be a similar colour, I let them stay as more individual but kept the orange over lay. As you can see I also tried adding some light to the buildings as well as shadows according to the sky. In comparison to my first matte painting, I do feel I have improved a lot. Instead of merely slightly editing the images, I have a better thought process and a better understanding of how to use Photoshop tools. I still feel I have a long way to go until I become good at matte painting but the fact I have definitely improved is a good progress.
I also wanted to add that in some of the matte paintings, the sky may seem pixelated but it’s only used as a template to let me know what the sky will look like in the scene. The v ray dome will be much higher resolution.
I imported the matte painting into the scene again and here you can see what it looks like. However, this is without the sky.
Deon helped me add the v ray sky and as you can see here, it looks a LOT better. The matte painting seems more realistic because it has taken on the sky’s lighting and it blends better with the model (despite it not having textures yet). Over all I am pleased with the outcome. It is a major improvement from my first matte painting test.
To help Deon with the next set of matte paintings, I decided to make a collage to help inspire him. These were taken from Tron Legacy, Mass Effect 3 (from actual trailer, game or concept) and Lost Odyssey. The reason why I chose these as influences is because they all have a futuristic look and an emphasis on technology. As you can see, the cities aren’t simple designs, there’s usually spires and complex looking textures.
Unfortunately, Deon didn’t have time to make more matte paintings, instead Gerome duplicated the matte paintings I did so give the illusion of many buildings in the horizon.
It’s a shame I didn’t use the projection techniques that I learnt from when I looked at matte paintings for the very first time. However, this would’ve been impractical due to the fact that the camera is very erratic. Projecting the matte paintings onto 3d objects would only work with subtle camera movements. This is why we used planes instead.
Also when you said the pivots centred, where should the pivots be? Do you mean pivots centred of each control, or the whole mesh? I’m not sure what you meant. Would you suggest I go back to the beginning to make sure I sort out these pivots first since I’ve already added orient constraint.
Third Year Final Major Project 