Showreel:

photoshop + illustrator

During the second experimentation I carried for the painting style for the background I thought I could import the keyframes into Illustrator after I had painted them an apply an image trace effect to them: the Image Trace feature in Adobe Illustrator is a quick way to convert your image to vector format for high quality printing at any size.

keyframe with the image trace effect applied to it.

This week I also rendered more keyframes to make the video longer (72 frames)

I painted the first key and every ten keyframe from there so that the transition would be smooth for the final output.

Ebsynth processes the input keyframes translating ten frame before and after those provided: this time I have placed the sequences according to the keyframes order in after effect setting an opacity to them so that the frame transition would look smoother.

render grey model, wireframe and final render

For the project submission thought that other than the ebsyth video I could render a wireframe of the model and the scene as well as the beauty and the paint effect to showcase the process behind the final result.

I have found this video to create the wireframe surface

I first applied a standard surface to it and created an aiwirame node in maya that I later linked to the standard surface base colour. I also changed the fill colour and the line colour to make them purple.

For editing I have used after effects and on each master layer I have applied some effects on the beauty layer such as the exposure and hue and saturation. I also overlaid the ao layer on top to give some depth and definition to the shadow as well. to the ao layer I then applied a gamma effect to edit the rgb channel.

ebsynth resolution pictures

In order for ebsynth to render the video the picture should not be in an exr format which I firstly exported the picture to. So after I rendered them as jpg the painted keys and the keys from the render would no longer match in resolution and ebsynth was not able to process them and kept giving me a “resolution error”. So I tried to manually make the resolution of the painted keys to match those of the footage in illustrator. Once I did that the program was working again.

submission and showreel

the initial version of the showreel had some presentation errors: the font I had used was too big and wrong, some of the bed renders appeared a bit too overwhelming since I had animated not only the camera but the model itself; moreover the resolution of the footage was not tv safe. So I have created a second version of the showreel applying these corrections.

I also used a different wireframe process since the other one was not correct as it appeared way too dense in the blanket and pillows areas.

I have applied a black surface and put the bed into a new layer changing the colour to white I also set the shader to wire frame.

this is a frame of the final wireframe shader I have used.

Once all the textured were imported and edited I have rendered the scene.

in the render layer settings I have created a bed layer a background layer and a shadow matte as well. For the bed render layer I have created two different collections one with the bed model and a second one with the background but with the primary visibility turned off so that the shadows would still be displayed on the bed model but the background won’t be visible. I did this since that afterwards I am going to manually paint on the background to add a brush strokes effect that I will apply to the footage using absinth.

This is the first experimentation I did painting the background in photoshop: I have set three different layers the bottom one being the rendered frame from the background I would the created a painted layer on top of it and at the top the png of the bed. I have chosen to not paint the bed model since I thought it would make all the work I did in substance painter pointless covering all the texture.

After I was done with painting stage of the frame I have imported the frame and the rendered footage from the master layer in ebsynth to apply the paint effect to the rest of the layers.

Although the frame was just one and the program did not know how to fill the parts of the footage that were not visible in the painted keys at first but were in the frame in the last part of the video ( the top of the bed for instance).

Afterwards I have painted two additional frames to add to the footage that were far apart and covered everything showing throughout the footage.

this is a screenshot from the ebsynth settings: there are two directories where you can import the original footage in a sequence of images and the painted keyframe. based on the directory of the folders with the images the program would automatically create a project directory for the output footage which would be in image sequence as well. The name of the keyframes chosen to edit should be the same of those from the original footage.

this output which I composited in after effects was not working as well: the images were not enough and the change in effect was to sudden the transition had to be smoother.

After I have created an additional frame and added it into the program.

This is the final output for the first experimentation:

Generators are substances that generate a mask or a material based on the mesh topology using the additional maps setup in the TextureSet Settings. Each generator has a set of parameters allowing you to fine-tune the resulting mask. Based on baked map information, they generate masks that can then be used to blend together materials and other effects.

I have added a generator layer to the different texture to experiment with the look of the model. I have used one of the preset “dripping rust” and placed it on top of the paint mask layers.

Although for the blanket texture it was a bit overwhelming so I did not use it.

As it was before

Baking the textures

After finalising the textures I have baked them before exporting them. Baking is the process of saving information from a 3D mesh, to a texture file (bitmap).

To create the blanket model and the pillows models I have used the cloth simulation in Maya. When I exported the model as an fbx to texture it in substance painter the blanket mesh was exported both withe the simulations applied and as simple planes. When I worked in substance I was able to hide it since it appeared as an independent texture set in the texture set list. Although, when I baked the texture the ambient occlusion appeared like this:

After I realised that what was casting the shadows on the blanket and the pillows was the plane that I had previously hidden.

By turning the ambient occlusion setting lower for the blanket and the pillows in the baking setting, I was able to solve the issue.

After I baked the textures I exported the different texture files following the guidelines in these two videos:

I have used the Arnorld output templates since I would have to import them in maya afterwards.

In order to apply the textures to the material using the right settings I have followed the guidelines provided by this table.

After I have imported all the texture files I have set the scene to use to give the bed model some context: I have created a wall and a floor following the painting room set up as well as creating a window from where the light comes in adding some green brush strokes on the glass. I have created a skydive to light the scene as well.

the problem with the first substance painter file

When experimenting with the bed texture files, I initially added the height, roughness and colour information on one paint layer for each texture I had created which meant that it was impossible for me to edit the height information, for instance, afterwards.

I will consequently use a non-destructive process. If your entire workflow is non-destructive, you can make edits way upstream in your process, and have the changes automatically propagate downstream.  

smoothed mesh

Moreover, the model I had imported in substance painter was unsmoothed which meant that when I would have imported all the different texture files in maya they would have look deformed on the smoothed object. So before starting working again on the textures I have smoothed the mesh in maya (saving a file where the mesh preview was still available) and imported it into substance.

layer process and height information

The new process I have followed included creating a fill layer with the roughness and colour information and a mask attached to it with a paint layer where I have created the different paint brushes.

This time I also inserted the painting into a colour picker program to select the exact colours from the painting into substance.

This is the texture render so far:

This week I have started texturing the bed model in substance painter.

texture sets and materials

In substance painter a texture set represents the material you assign to mesh in your 3D application. Each individual material becomes a texture set with its own dedicated layer stack in the channels. When you export each texture set results in a set of bitmap files with a file for each channel of the set. Texture sets can be enabled and disabled in the texture set menu using the eye icon. Each texture has its own layer stack

materials and channels

Material can be broken down into individual channels each channel represents a material property: base color, metallic and roughness. Fill layers can act as a material layer stack as they can fill one or more channels with a texture or a uniform value.

baking

Baking is a step where you save or bake geometry information from your mesh into texture maps. Baking can speed up the process a lot. The information stored into the maps is used to create procedural effects and it can even help with your material placement.

paintable layer and fill layer

Paint Layer : This type of layer can be painted on with brushes and particles.

Fill layer : This layer cannot be painted on, instead you can load a material into it, to fill the channels. (You can also manipulate the transformation to repeat the material for example.)

Van Gogh painting style and the 3D model

Van Gogh is well known for his brushstokes of thickly laid-on paint. This technique is called Impasto. An artist lays a thick layer of paint on canvas, brushstrokes get more noticeable, adding a special texture to the painting. Vincent liked to use a thick, undiluted flat color with a brush or a palette knife.

This following video describes in depth the process of Van Gogh style evolution over the years. It was very interesting to see what inspired his work and how he developed his own “voice”. During the presentation the simultaneous contrast is mentioned in regards of the painter artworks: it refers to the way in which two different colours affect each other. The theory is that one color can change how we perceive the tone and hue of another when the two are placed side by side. The actual colors themselves don’t change, but we see them as altered.

This is a close up picture from the “bedroom in arles” painting. As it showed the paint is very thick and dense in some parts which is what I am would like to render in the 3D texture.

I first imported the bed model as a fbx file into substance painter: substance painter recognise the different texture attributes created previously in maya, so I have assigned to the different parts of the model different textures according to their material

I first applied a base colour according to the colours of the painting itself

I after started working on the bed wood paint texture using an alpha material with a paint brush to simulate the brush strokes. I also included some height information to simulate the impasto technique used by the painter.

I have first experimented with different brushes to see the look and the amount of detail I was able to reproduce.

This first attempt was a bit coarse and the dark brown edges I had created were too dark and resulted in a “fake” outlook. I first created a layer with light yellow brush strokes and after created one on top with a light red: I tried to pick the colours according to those in the painting

Using a similar process I started again but this time using a different brush and height information. I kept the dark edges at first but they were to predominant in the texture.

I after smudged the edges and added a light desaturated green to them as well.

I also started texturing the blanket, the pillows and the mattress.

UV map

Once the model was ready I started to Uv map it. UV mapping helps you locate your image map precisely on your model. UV refers to coordinates that are independent of the XYZ axes. To work with UV maps, you need to be able to unfold and fold your 3D artwork, and also cut and sew parts together.

The approach I have used is the following:

For each part of the model I first applied a planar projection (Planar mapping projects UVs onto a mesh through a plane). When a model’s form is very organic, and you require a complete UV representation of that model for texture mapping, a planar projection can produce UVs that overlap and appear distorted. So I unfolded the UVs to make them usable.

Once I have planar projected the individual meshes before unfolding them I have used the cut tool. Cuts are an important part of the unfolding process because they allow the UV texture coordinates to freely unfold with as little distortion as possible. When unfolded, the UV mesh should lay flat without overlapping.

For the initial stage I have applied a checker texture to each mesh: the checker pattern shader is applied to the surface of your UV mesh and appears behind the grid in the UV Editor. You can toggle between a simple black and white shader to a color gradient in the Checker Map options. The color gradient makes it easy to locate tiles on your UV mesh.

For the bed head I actually planar projected the different faces according to which axes they were belonging to (x,y or z)

After I have arranged the UV layout of multiple selected objects simultaneously according to their material: so in this case I have separated the wood parts from the pillows, the mattress and the covers so that when I will texture them in substance painter I will texture them separately.

For the rectangle shapes of the bed heads I have edited the the cuts since the way I had initially cut them and stitched together were not ideal for the texturing stage: by placing the cut at the top (except for one edge), at the bottom of the geometry (except for one side) with just one cut on the side edges, all the cut are placed in areas where will not be seen according to the way the geometry is modelled which will leave any discontinuance hidden.

This week I have completed the modelling.

I started from adding some edge loops on the shapes since that when switching from low and high resolution view the model subdivisions where collapsing and also to make the subdivisions to be evenly distributed on each part of the model

After

Applied to the whole model:

This week I also continued experimenting with the blanket model to position on top of the mattress

I started from editing with the cloth I had created in week 12 and used the lattice tool: it surrounds an object with a box-like wireframe structure; this is the lattice the deformer is named for. The points of this structure can be selected and transformed, which also transforms the portion of the surface they enclose.

I also tried to apply the cloth to a squashed tube polygon to see how it would react compared to the plane, although the subdivisions were not enough and it was not folding enough.

After wards I carried different experiments where I have deformed the plane before applying the ncloth to it such as the soft selection to move the vertices of the plane subtly with though ruining the subdivisions levels

I ended up using a mattress cover, since I have noticed that the bed in the painting has one, and adding a deformed plane (on which I have applied the cloth to it) on top. This time I have also added as a passive collider the bed head so that the plane would bounce on it and create creases.

This is the final model:

This week I have started to model the bed from the painting for the project

I have started by importing the image of the painting as a reference

I have started by modelling the bed head starting from a cube adding some subdivision levels and selected the vertices to shape it. In the whole process I have switched between low and high resolution (1,3 key) to check that the volume was not collapsing.

In order to add volume to it I have extruded some faces on both sides

After I have mirrored the bed head and made it bigger for the other part which would be the other corresponding to where the pillows are.

To create the blanket and the pillows I have followed this videos tutorials that I have found on youtube and took inspiration from them.

I also created the side of the bed from a cube and used the insert edge loop to add subdivisions

I also bevelled the corners to made rounder corners and mirrored it to the other side

I after created the the mattress by scaling up a cube and bevelling the edges of the corners

I also started experimenting with the blanket shape and model using the nCloth simulation: simulates a wide variety of dynamic polygon surfaces. For example, nCloth is flexible enough to simulate all the following surfaces: fabric clothing, inflating balloons, shattering surfaces, and deformable objects.

So I have created a plane increased the subdivisions levels and added the cloth simulation to it. I have also created a passive colliders from the bed polygons mesh in my scene and have it passively interact with nCloth. I played the simulation with the plane starting from a tilted position on top of the bed and waited for it to fall on it and decide the shape. I after duplicated it and deleted the history of it to keep the final shape.

For the pillows I have scaled a cube very thin and increased the subdivisions levels and applied the cloth to it. This time I edited the gravity of the effect to be very low so that it would not collapse and played the simulation on the timeline choose a key where the mesh had a “natural” pillow shape, I duplicated it, deleted the history and positioned it on the bed.

Initial stage of the whole model:

Modelling/sculpting and texturing one asset from Van Gogh’s painting “Bedroom in Arles” as a case study

For my FMP I wanted to create a short 3D animated film revolving around Van Gogh: my idea is to create a Van Gogh avatar which would take the viewers on a tour of his most famous paintings, which I would create a 3D version. I am planning to create a 3D “house” where the elements of the different painting are a part of.

According to this idea, for this premise project I want to start and create the 3D version of an object from Van Gogh’s “Bedroom in Arles” since it would be the focus of the animation and I would be able to concentrate more on the details of the texture, modelling and UV mapping which I will after apply on a bigger scale. It would also be challenging for me since Modelling, Sculpting and Texturing are skills that I am still developing.

This project would also be useful for me since I am planning onto using substance surface for the texture and this is a software I have never used before.

The asset I am going to model is the bed, since is one of the asset which draws the attention in the painting.

The following are some concepts ideas to use as a reference for the project: