These are the 4 final images for the assignment of the fruitbowl.
Image of what we were given to render for comparison
This year 2 subjects we took were rendering and texturing. ‘Texturing (is) the process of applying colours and textures to an object to make it renderable’ (Derakshani, 2011, p 9) Rendering is the process the computer uses to calculate lighting, shadows, shapes of objects and movement and saves it in a sequence of images to produce as realistic a result as you want. (Derakshani, 2011, p 495)
Texturing was good to learn because I had a bit of trouble with it last year. This semester it seemed quite hard to learn what seemed vast amounts of knowledge poured out in the class but that was understandable because we learnt how to use 2 whole new programmes, UV Layout and Mudbox, and there were only about 4 or 5 weeks to get all the knowledge. With both the programmes I felt panicky like it was too hard and I could not do it however with patient help from Eric and lots of practice for the assignments I got quite good at them.
UVLayout is used for UV texture mapping. UV’s create a space to paint textures on models. Texture mapping is putting 2D images on 3D objects to add colour and detail. It is a necessary part of completing a model. (Honn, 2010, p. 573)
UVLayout creates maps easily, with little distortion, which are easy to paint on in Mudbox. It was easy to understand but with the whole programme to learn in 1 lesson it was overwhelming. However there were videos to watch which explained everything and I got the hang of it. UV Layout makes it easy to see where there is distortion by different colours for stretched, pinched and relaxed areas and it has an automatic flattening function. We had to use this progamme to UV map Blobby Boy and I like how much easier than Maya it is to use once you know how. In the end I believe I got good results UV mapping Blobby and there was not much stretching on the whole map,very little blue or red shading on him.
The other texturing programme we learnt was Mudbox, a programme for sculpting and painting an object. It is good for painting the UV map while it is still geometry so that there are no real seams to try to match. It allows you to paint with stamps and through stencils and use images that you have created in photoshop or even photos with the projection paint tool. You can create maps for ambient occlusion otherwise called a dirt map and normal map to plug in to the bump node of a shader.
At first I could not get how to use it either alone or in conjunction with Photoshop and Maya. Eric was a great help with this and it is not too hard after all. It is now easy for me to transport images, maps and geometry between the 3 programmes. Now I am quite proficient with it.
The other half of texturing was Photshop with Lilly. Photoshops speciality is painting and creating images. This was an enjoyable class and I learnt a lot of new things like how to use a tablet and pen, paint in layers and paint images of what we were looking at. Coming from an artistic background of drawing, oil and water colour painting, my pictures looked like paintings rather than realistic photos. We learnt how to use adjustment layers and things like transparency, multiplying layers, colour burn and overlay. I found this very useful especially in how to use layers in mudbox and in making swatches for the fruit bowl.
We also learnt how to use a photo and make a brush from parts of it to paint an image of a surface, say a tree bark or rust, which I used to make brushes and textures for the fruitbowl assignment.
Rendering was a whole new subject. We learnt all about Mental Ray and things like anti-aliasing, final gather, caustics, mental ray shaders, sub surface scatter, motion, blur, depth of field, ambient occlusion, physical sun and sky, mental ray shadows, render passes, colour balance etc.
Mental Ray is a much better renderer than the Maya renderer. It can create photo realistic renders.
Anti-aliasing is critical in rendering. Without it you can get jagged or stair-stepped artifacts along diagonal edges or where there are a lot of small objects or lines together. (Birn, 2006, p. 259)
Global Illumination is wonderful, the programme computes the bounces of light between surfaces and the colour bleed that happens, it replaces fill lights and their shadows (Birn, 2006, p. 74)
We had to render 3 objects to look the same as 3 real world objects and I really like how my gold shiny christmas bauble looked but I forgot to adjust the samples on the glossiness for my pink plastic elephant watering can and the shine was too grainy.
Finally at the end of this semester we were given a few weeks to texture and render a scene in Maya using all we have learnt in Rendering and Texturing to produce realistic looking fruit, light and a glass bottle of water in 2 scenes with 2 lighting scenarios. It is such a huge assignment with many blog posts required as well as just doing the thing itself, I hope I can do it all in time by 5pm tomorrow!
I had lots of trouble at the start trying to figure out how to use Mudbox and the pipelines between the programmes. I had some difficulty after making what I thought was a good sculpted texture Mudbox to show up in a render but it was the size of map I was exporting 32 bit and 16 bit floating had too much information and it didn’t work until I made a 16 bit or 8 bit integer map. It takes me a very long time to get the renders I am happy with starting from making brushes and textures and maps to placing lights and cameras and working out the best render settings and waiting for each test render to show results before twiddling things and trying again.
Overall I have learnt heaps in these subjects. It seemed too much to learn but it is possible after all. UVLayout, Mudbox, Photoshop and Mental Ray are now tools I can use in the pursuit of excellence in texturing and rendering.
Birn, J., 2006. Digital Lighting and Rendering. second ed. Berkeley(California): New Riders.
Derakshani, D., 2011. Introducing Autodesk Maya 2012. 2012 ed. Inianapolis(Indiana): Wiley Publishing.
Honn, E. K. w. P. a., 2010. Mastering Autodesk Maya2011. 2011 ed. Indianapolis(Indiana): Wiley Publishing.
I made many textures swatches and not all of them are in the final result but all were tried and the best were kept.
I used brushes that I made in Photoshop to create 2K square images with white marks on transparent to sculpt and paint through in Mudbox, and to paint with in Photoshop to create other swatches to plug straight in like for the banana (see below).
brushes I made and used to create swatches and some of the swatches
I had a look at the book ‘digital lighting and texturing’ about textures and it talked about photographing things to use as textures and gave an example of banana skins (Birn, 2006) . It said to get no shine or shadows and make the surface as flat as possible, then to clone the skin in parts between the actual skin sections.
I first tried the only bananas we had which were old and nearly black but because I scanned them there were wrinkles and it didn’t look that good. I also have to make the ends and the butt different to the rest of the skin.
I then went and bought some new bananas and peeled one and photographed it and photoshopped it, making swatchess to use for diffuse and bump. When I put the texture on all the bananas they all looked exactly the same so I used a few different banana textures that I made but they were too different for a bunch so I used the same one and changed the placing of the uv maps on the texture so that they are similar but different. When I tried to export the bananas as obj’s to work on it in Mudbox it kept not working so I did it in photoshop.
This is the steps of the texture I did in photoshop
I made a tiling stencil in Mudbox using a swatch of white holes on transparent made in photoshop from a brush I had made in Photoshop. I made small and larger holes and sculpted through it into high subdivision levels of the orange, and made a good skin for an orange. I had lots of trouble getting it to look good in Maya but eventually new sculpting and painting through the stencils I made yielded good results.
Some of the swatches made for the orange and the results.
I tried many different things for the apple texture. I painted in Mudbox using transparent stencils that I made in Photoshop and using colours from the photos. But I was never really happy until I made a square image from a part of the skin of a red apple cloning and patching til it filled the square. I used a brush I had made of the dots of an apple eye-dropping the colour from the dots and painted in more where it got too plain.
I plugged that in to the diffuse part of the mia shader but noticed that the stalk was red too. Then I saved a copy of the uv map and in Photoshop lay that over the red square and made it semi-transparent, then painted the stalk and bottom with a colour eye-droppered from a photo of an apple stalk.
I painted some pale green from the photo too in patches, semi-transparent and it started to look good. I changed the file colour balance to be more green but when I added the sticker to the file in photoshop that too was green so I then did the greening in photoshop apart from the sticker.
I made the green apple by painting layers of colours in mudbox and used the dots swatches and made a few layers of them too, adjusting transparency and erasing parts made a good result that I am happy with.
I tried a number of times to make a good texture for the pears and at last I am happyish. The first lot looked like avacados. I started again in Mudbox and painted lots of layers of colours and erased some patches to let the layers beneath show through. I made layers of different coloured and sized dots that I made in a tile in Photoshop, white on transparent, to paint through and also to sculpt through. When I put it on the pear I wasn’t too happy with the colour and altered the colour gain of the image trying greener, yellower and finally being happy with pinker. Here I have some of the variations of pear colour and some of the tiles and stencils I made in Photoshop.
It was time to do the banana ends. By now I am very comfortable with Mudbox and decided to paint 1 with a projection of the banana skin tile that I had made and then paint the sides dark which were dark on the real bananas I had. This worked well so I decided to try to repaint a banana too to make the ends darker to match. I painted the projection again and then painted the ends darker too. I loved it and made a copy and made that grey scale to make a bump map and again to make a specular.
I then remembered the fun I have been having with reflections and didn’t use that. I changed the bump layer to a normal layer and plugged this and the colour into the shader. The colour looked good but the normal made it terrible. Then I remembered to change the colour profile on the image to linear sRGB and the bump tangent space normals. I thought the result still too lumpy so I tried saving a more transparent layer as the normal but there was not much difference so I plugged in the bump map.
I think the Photoshop texture looked better overall than painting it in Mudbox, 1 reason is that painting the projection looks good til it goes on the side around the corner then you have to paint over the stretched dots and it can have a different colour if the place on the tile is not right. The difference in the photos is more obvious if you click on the photo and look at the bigger photos.
In Lilly’s class we made a texture of wood by looking at a piece of wood and trying to get the colours and patterns right by eye. I lightened it a bit in Photoshop to make it more like the references given for this assignment. I have used it on the wood of the fruit bowl and I like how it looks. I don’t mind how it has seams because it looks like real wood cut and stuck together.
I used photos of a stone wall and made it a 2K square tilable image. I used this by painting through the projection of the image over the tiles but when I plugged it in to the diffuse and bump ( as either a bump or normal map) it didn’t look good so I took out the bump and made it shiny tiles but it still wasn’t good enough because there was a deep looking bump in the pattern.
So I took an image from a beautiful old Napoleonic bridge in Paris and merged that with a stone looking pattern I made with my own brush and colours from the bridge. It looked great in Mudbox but just doesn’t work well at all in Maya, the bumps and reflections aren’t good enough and the pattern doesn’t show up enough. I finally painted a new one in Photoshop not worrying about it being able to tile because in the real world tiles patterns are often repeated, I copied and pasted the image the same number of time as tiles in the geometry, turning some 90 and 180 degrees, and painted off white where grouting would be using layers. I used a painting of granite that I made in Mudbox for the benchtop.
Birn, J., 2006. digital Lighting and Rendering. Berkeley CA: New Riders.
I am on the final run. I have assigned textures to everything.
I have been trying out lights, eg whether to use physical sun and sky or normal lights.
Weakness– I had put in lights at the start but light linking the caustics and grapes lights may be what stuffed up my renders and shadow from the window. Nothing worked so with Chris’s help I deleted all the geometry and imported an earlier one now the window has a shadow with a directional light.
Then I discovered that every spot light I made for outside the window had the same trouble.
Strength- So I made a directional light and changed the light angle to blur it up a bit.
Strength- I am putting in ambient occlusion on the shader nodes to have more of so that it all looks more real with that and I like the way that looks.
Weakness- I am fiddling with the grapes subsurface scatter node’s specular layer and am discovering that I can make the shine look more like that on real grapes, more diffuse. But I am still not happy with the look, I want it to be more cloudy on the surface.
I still want the lighting to look realistic. I am getting either too bright or too dark with the phyiscal sun and sky so that may have to go.
I want good caustics too.
I am having heaps of trouble trying to work with Mudbox and the pipelines between it and Maya and Photoshop. However I made a tiling stencil in Mudbox and made a good skin for an orange. I managed to make maps for ambient occlusion and displacement. I plugged them in to the mia shader in Maya in diffuse weight and bump and the look is way over the top. Every step of the way takes me so long because I am learning as I go and trying everything.
Here is what I have at the moment. It needs shallower pores and I should do the map again with no blurriness. That came from sculpting through the stencil on too low a level in Mudbox and doing it on the side at an angle where it is blurry. I realised later that I had forgotten to make the grid larger by 1,000 in maya and to make a huge copy to export because the scale is bigger in mudbox.
I redid the orange and maps in Mudbox got better maps and plugged them in to colour weight and bump and fiddled with reflection and gloss; and painted a colour in Photoshop, which I plugged into colour in the mia shader. I am much happier with the result. However it needs better pores and more ambient occlusion.
The lights for the grapes are casting light where I don’t want it so I am light linking a back light and a front light to them and seeing if that looks better.
Not too sure what to do wtih a displacement map or AO map, and not getting good results in my renders.
Today, Tuesday, we learnt how and where to plug in a displacement map into a mia material x shader so now the orange can look better than at first when it went into the bump part. You drag it into the shader node instead, there is a dedicated part called displacement map.
Today, Wednesday, Eric came and helped me with all the problems I am having with Mudbox. I learnt we only need a normal map and an AO, or dirt, map for Maya for this assignment, the other maps displacement and vector displacement are for modelling. Also how to sculpt the surface for blemishes and creases and that these will be saved in the normals map and plugged in to the bump node in Maya. He also taught me how to find and use the painting that I do in Mudbox and how to paint a grey scale image over blemishes for the reflection part of a maya mia shader so that these parts will be less shiny, and to paint in dirt in crevices for the dirt map. Also I learnt that we make a stencil in Photoshop like Lilly taught us and how to bring it in to mudbox and that we can use that as a stencil or stamp.
Friday- time to practice what I learnt… it takes ages to get the results I want. I am redoing the orange and bananas in Mudbox trying to get them better looking. I started in Photoshop and made new stencils. I also tried the mia skin shader on the orange to see if I can give it a more realistic look with the depths. That didn’t work well. So it is back to the mia_material _x.
Made a better stencil for the orange in Photoshop, imported a huge orange to Mudbox and subdivided it much more and made bigger deeper pores with the new stencil and creases, hoping the new normals and ambient occlusion map work much better. At this rate I am not sure I will finish it.
Spent days and days of stuggling with the orange and frustratingly finding that absolutely nothing worked.
The problem was figured out with Eric. 32 and 16 bit floating settings when extracting the normal map in Mudbox meant there was too big a range for the renderer to find the information and the result was always too different, flatter than what I had sculpted in Mudbox. We experimented with these settings and found that 16 bit integer was much better but hardly any better than 8 bit integer so finally with this setting I have an orange.
Strength- it look like an orange now
Whenever I have painted a specular layer in Mudbox for a reflection map by painting black where I didn’t want reflections the result, when I plugged it into reflection in Maya, was that the whole thing became as shiny as a mirror.
So I looked at the reflection value that I had chosen in Maya with out the plug in and painted that value of grey on the layer in Mudbox where I did want some reflection, say .03, the result was always the same a shiny mirror surface, even where I painted a value of 0.
It is very frustrating so I left it out and just used the low reflection value for the whole thing. They looked pretty good in a scene on their own but back in the original scene it’s a disaster.
Below is an example of the very shiny apples which have been causing me grief.
The avacado looking pear and the speckles which will hopefully be fixed later with rendering settings, like samples.
After adjusting the reflection I am happy with the apples.
I didn’t like how it was angular and tried a round corners mia thingy but it didn’t look right.
Strength Then I plugged a displacement node in to the shader group and immediately it looked more normal in the render.
This is based on one of the photos we were given with outside light coming from the right, here through a window, as in late afternoon or early morning. I may colour the light, trying pink or yellow, to make it more realistic for morning or afternoon.
Strength- This screen grab of a render has no GI or FG and low sample rates to quicken the render time, but I am happy with the way the fruit and the wooden basket look now.
Weakness- Grapes still not good enough- The lighting behind the grapes is a bit wrong with more light on the left away from the light source than the right. the lights may be too close and causing uneven patches of light.
Desire- A desire that I have is to have the window cast a shadow but I can’t make it do it yet.
Desires- The pears and grapes could look better. I want my grapes to look more real with maybe some cloudiness on the surface. I have tried green pears and brown pears.
I have to finish getting good lighting and setting up the scenes
I have to do good caustics yet. I want concentrated refractions of light coming from the bottle.
I want good realistic shadows.
We have been given a model of fruit in a bowl and a bottle, to texture and render as life-like as possible. I am looking at references like the photos and real fruit and bottles as I do anything.
I started by placing lights to light the scene and some cameras for the 2 views but as I apply shaders I am experimenting with lights and camera angles for the best results. I attached a measuring tool to the lens of a camera, using the node editor I plugged the distance of the measurer into the depth of field of the camera so the focus length is always the same as the far end of the measurer tool and I place that where I want the camera to focus. This is good for close focus with things behind and in front out of focus.
I applied mia_round_corners shader to the grapes stalk to make it rounded in the render view rather than angular as the model is. I also put it on the wood so that the corners on the wood would not be so unrealistically sharp.
The grapes have sub surface scatter so I applied a misss_fast_simple_maya shader to the whole lot of them, 191 grapes. The grapes are not exactly the same colour so I applied a randomised diffuse with a mel script written by a teacher. Now they are all differing according the the different green colours I put in the ramp. Varying the values on the shader with light behind and in front, it is quite difficult to get them looking like grapes and not olives or lights. Also they look quite granular so I think I have to improve anti-aliasing and samples.
I have started the bottle of water. I made a light for it with intensity at 0 and made it emit photons only, caustics after turning on caustics in render settings, indirect lighting tab. I made the light face the bottle and enclose it in the cone angle. I am in the process of fiddling with the photon settings of the light to get good refractions or caustics. I increased ray tracing samples for reflections and refractions on the quality tab and same for the photon tracing on the indirect lighting tab. I reversed the normals of the water sides and top inside the bottle and applied mia_material_x shaders to them. I am very happy with how real it looks but I think I need to do more to get caustics showing on the bench and also the bottom of the bottle needs more attention I think.
Weaknesses- The grapes don’t look real they look like lights, I need to spend more time on them to stop them looking like plastic lights. They are granular too. The stalk doesn’t look right with the rounded shader on it and I will have to discover a better way to make it not be angular. The caustics are not right yet. I want to see the bottom of the bottle too.
Strengths- I like how the grapes have real differing green shades and there is subsurface scatter going on. The depth of field is working nicely. The bottle of water is transparent and looks like what it is meant to be.
Desires- I want the grapes to have the right amount of subsurface scatter so that they look like grapes not plastic or olives. I want the caustics to shine sharply on the surfaces around and to see the bottom of the bottle too.
I saved the blobby guy as an obj file and then dragged it into UvLayout
I cut and split in places which were least visible like back of the head, inside arms and legs and around the edges of the clothes .
I made strategic cuts up the back of the head to the forehead and around the mid-line at the back to the side so that the seams would not be obvious on the face, and so that it could be flattened with the least amount of stretching. I cut around the eyelids and lips and base of the antennae, areas which have a lot of in-folding or extend out. I cut the uvula, or the thing that hangs at the back of the mouth, down the back and along the middle of the bottom to minimise seams just in case he opens his mouth in a close up.
I cut the body where seams would go in clothes, and around where the legs and arms join. The clothing here bends under out of view. I cut the shorts and shirt where clothing seams would go. I cut down the inside of the legs and along towards the front on the foot and put 1 seam down the inside of the arms where they would not show obviously and unwrapped them in 1 piece each.
I cut the hands along the thumb side but towards the palm a bit and around between the fingers leaving the other side if the hand intact, then split it.
hands laid out
Doing the face was complicated because it is usually a focus of attention and there are small details. I made the edges at the back of the head line up which is good for painting in Photoshop. I changed my mind later to make it less distorted for Mudbox instead- see later. I made the bottom edge straight to aid in keeping the mouth open when it unwrapped but later decided less deformation is much better and that it didn’t matter at that edge because it ends inside the shirt anyway.
whole body cut and laid out
For Photoshop edges have to be straight so on the arm I anchored and pinned the corners. I lined up edges with an anchor point, it is stretched but it is straight.
I fitted the mouth and eyelids next. I moved the chin down to the same rounded curve as in 3D view and pinned it to allow room for the lips to fit in the mouth hole and B brushed it. I fitted the eyelids and mouth by rescaling to local and manipulating uvs to generally the right shape and size then welding in place. It is so there is no seam around the lips or eyes or the nose hole. I fitted in the bottom row for inside the nose only because that is all you see and it needs to have no seam there. .
Altering the lines in the face around the eyes made more deformation which I am finding hard to correct.
I flattened the legs like the arms into a square to paint in Photoshop .
I straightened that edges at the end of the hands to match up to the arms at the wrist which again is good for painting in Photoshop which needs straight edges. I decided later to make everything have less distortion to paint in Mudbox.
The hand and arm line up
I set bleed at 4 to pack it to get an 8 pixel gap between shells so they aren’t too close or overlapping at all and so I can paint easily on each shell without painting on the next 1. Also so I am not wasting space.
I decided to split the pants so that seams could go where real ones do up the front but later decided against that because less seams are better and you can paint seams and zips in.
I finally decided to uv map for painting in a 3D programme like Mudbox which means less distortion is more important than straight edges like for a 2D programme like Photoshop. Because of this I changed the squared off shapes to ones with less distortion eg the arms and legs and the neck of the face.
I redid the head by pinning the outer corners far out and letting it stretch which allows room for the face to open out. Then I pinned the face in place moved the outer pins and allowed the rest to relax more resulting in less distortion over all.
I moved the seam of the shorts to the back to be less visible and welded the tops of the shoulders too again for less seams.
The arm and hand are joined so I can paint straight over the wrist with no seam. The top of the arm edge doesn’t matter because it is under the shirt, same with the top of the legs.
I redid the face and like it better. I handled overlapping places like the mouth and eyes with the overlapping process. (I don’t know why the lines went yellow)
There is little distortion over the whole character, overall it looks good.