Marco Oliveri | Texturing Artist / Generalist

Welcome to my latest project, the Yashica FX-3 Super 2000.

I took my parents’ old analogue reflex and I decided to recreate it in 3D starting from scratch. It's been a long journey, quite challenging for some aspects since I had never modeled anything this complex before.

Photogrammetry

Since I wanted to model the reflex as accurately as possible, I decided to do a photogrammetry of the physical object. It was my first time approaching this kind of procedure so I looked for some information before and then I took all the necessary. To keep things simple, I built a lightbox from an old cardboard box I had at home and I covered the inside with a big paper sheet in order to create a limbo, then I crafted a simple turntable and I positioned the reflex at the center of it.

Dealing with the glossiness of the reflex surface has been challenging since it could have subsequently compromised the photogrammetry calculation. Reading some forums online, I found out that spraying talc powder on the glossy parts of the object could be a good way to fix this problem, so I tried and it worked surprisingly well.

After taking the photos, I used RealityCapture to elaborate them. It turned out to be a very efficient and great software and with a few guided steps I managed to obtain a pretty decent photogrammetry of the reflex. My goal was not to get a super accurate mesh, since I would have used it just to block out the proportions and volumes of the model.

This is the final result after the processing:

And below you can see the mesh imported on Maya:

Modeling

This has been the most time-consuming part of the whole project for sure. I spent lots of hours working on the modeling, trying to refine the shapes and get to the best possible result. I had already modeled hard-surface objects but this was way more complex. 

I took a bunch of photos to see all the details better, from the biggest shapes to the smallest elements. One thing I also chose to do during the modeling process (since I had the opportunity) was to measure the pieces of the reflex to better replicate all the distances, thicknesses and diameters of its parts. 

At the end of the modeling, I created the UVs and packed them in 10 UDIMs, to have more resolution later in texturing.

Texturing

Since I studied Foundry Mari during the previous months, I decided this to be the first project to test what I learnt. It was also my first time approaching the specular/glossiness workflow: it’s not been much intuitive at the beginning since I was used to the metallic/roughness process, so I had to look for more information about specular map and its values in order to match material physical properties in real life.

A big step in the texturing process has been recreating the various logos, signs, numbers and all the other elements on the reflex. I took many photos of all these details and then I processed them mainly with Photoshop and Substance Designer (I used the second one especially to extract and create scrapes and corrosion masks). Some writings were difficult to extract from photos so I created them from scratch using Illustrator.

After this, I generated the auxiliary maps on Mari using the new Bakery tool (personally, very powerful and fast) and I began to paint all the logos and signs using the previously created masks.

I'm a big fan of node-based software so I couldn't wait to try nodes workflow in texturing and it's been a pretty satisfying experience.

Dust Generator

Since I wanted to make the project more realistic, I decided to add dust on the model (also because the reflex had been on the desk during the whole project and dust started to settle on it).

After some tests with Maya MASH, I realized it was too restrictive for my purpose. Therefore, I exported the dust particles created with curves and I decided to simulate the effect on Houdini. I have to thank my friend Mario Leone who gave me some useful starting tips to setup the basic node tree, since I was completely new to Houdini and I didn't know how to proceed. After creating the base scatter setup, I spent some time working on it, improving the dispersion and size of dust, creating procedural masks based on position and ambient occlusion and then enriching them with an additional manually painted mask.

Finally, I exported the scattered particles from Houdini to Maya as a reference.

Rendering

As usual, I used V-Ray as my main render engine because it is fast and reliable (the reflex has been textured and rendered in ACES color space).

You can see below various turntables: beauty, lights, render passes, uv checker, wireframe and clay version.

I made this project to test my limits on procedural texturing and because I've always wanted to create textures for a kimono, too.

The kimono model was made by Nasim Beyt.

A little cinematic I created to present the kimono project (you can watch it in 4K, too!) 

References

The most important thing in such a complex project is to analyze all the elements in the reference and break them down in multiple groups to organize the work later in a more efficient way.

Shapes

Initially, I created all the elements as silhouettes and I proceeded to put them together inside a single graph to see how they worked when assembled and to have a settled node tree for subsequent iterations.

After this step, I elaborated the basic shapes into more detailed versions. Cranes have been for sure the ones which took more time to be crafted since they were all different and I wanted to depict as many details as possible. To do this, I had to create multiple masks inside Adobe Illustrator to separate the body parts from back feathers, beak and eye.

Cranes masks created in Illustrator

Evolution of cherry blossom and bush weaving patterns

Assembly

When all the elements were finished, I put them together in the initial graph and I generated all the other maps (base color, roughness and metallic).

Additional Fabrics

I also created two other fabrics (cotton and satin) to enrich and diversify the inner parts of the kimono.

Weaving scheme of satin (left) and cotton (right).

Details from finished textures

Rendering

I rendered the finished textured on Maya using V-Ray GPU and the turntable setup made by Cave Academy.

You can find more information in the article I wrote for The Rookies, where I gave a lot of insights about my workflow inside Substance Designer

This cinematic animation has been a great opportunity to approach fields of CGI in which I didn’t have much experience, especially some that I’m going to show you.

Cloth Simulation

The first big challenge was certainly the banner cloth simulation. I tried various approaches between Maya, Marvelous Designer and Blender but, in the end, I managed to get the desired result with the last one.

Despite the banner fluttering in a sinuous way, it was still moving too fast and I wanted to create some kind of slow-motion for my shots. After some trials and errors, I found out that I could use the point cache data export. This way, dynamic data of all simulation points (in this case, the vertices of the banner) for every frame of the simulation could be stored, so I got keyframes for all vertices that I could manipulate expanding them on a wider number of frames. This allowed me to slow down the cloth simulation.

After adjusting the UVs of the banner, I moved forward to the next big step.

Procedural Texturing

I created the fabric using Substance Designer to use all the power of proceduralism.

First, I started creating the subject of the banner, the three lions. Using a reference I found online, I traced the shapes with Adobe Illustrator and I then isolated multiple masks to use later within Substance.

At the same time, I realized the basic fabric which would have been applied to all the area of the banner. I went for a sort of velvety texture which I enriched later with a pattern of hand-sewn diagonal lines and fleur-de-lis (both these shapes and the lions were conceived of silver threads).

Evolution of fleur-de-lis

Evolution of velvety texture with decorated patterns

The lion textures have been a great challenge: I wanted to give them the feeling of being hand-sewn but also making them look sophisticated. That’s where the vector masks came in handy: I used them to isolate certain portions of the lion to achieve sections sewn at different angles and to place gemstones on teeth and eyes spots, too.

Evolution of lion weaving effect

After completing the height map, I created all the others and exported them to Blender.

Fur

Since the textures applied on the mesh looked a bit plain and I wanted to break up this effect, I decided to add a sort of coat of threads above all the surface.

I created a particle system and I scattered a curve across all the banner until I got the final look. This radically changed the perception of the material during the lighting and rendering process.

Lighting and Camera Layout

This was a big trial and error phase. I made a lot of tests, trying to achieve the most appealing camera movements. One of my main focal points was to use light to bring out the silver shininess of hand-sewn elements.

Rendering and Compositing

After reaching the desired outcome, I proceeded to render all the shots as EXR image sequence using Cycles with GPU (I chose to render everything inside Blender because there were a lot of frames and Cycles is a really fast engine).

I made all the compositing adjustments on Nuke: first of all, I shifted the color grading on colder and more radiant vibes, then I added some effects like blur, glow, chromatic aberration and also motion blur to blend the animation better.

Finally, I exported the shots to Davinci Resolve to assemble them and add the music (song credits to "ASTROFREQ" on Pixabay).

Shot compositing from raw render to final look

Thank you for your time!

I wanna thank whoever helped me throughout these projects, even just with a word or some advice, I think it’s fundamental to discuss and see someone else’s point of view, too.

I also thank all the people who took the time to read my entry and made it to the end. Please leave a comment, let me know what you think about my projects and if you have feedback for me!

You can reach out to me if you're interested in my works!

Here are my main social profiles:

Linkedin: https://www.linkedin.com/in/marco-oliveri-2b0959222/

Artstation: https://www.artstation.com/marco_oliveri

Instagram: https://www.instagram.com/marcooliveri_/