Interview: Gregory Smith
1. Tell us a little bit about yourself: Who are you, what do you do, and where are you located?
My name is Gregory Smith, and I live in England. At the moment, I’m in a weird intermediary year between a foundation year in fine art and a degree in CGI at Southampton Solent University. I’d always considered myself to be an oil painter, but started using Blender for making still images last year, and now prefer it to traditional methods. I’ve also been talking about starting some remote work with Barnstorm VFX.
2. Where did you find the inspiration for your latest entry?
Since I wanted to make a fantasy image for this topic, I went straight to ArtStation, and searched for ‘forest’ – I turned off ‘Show PRO member artwork first’ and filtered it by likes. Then I just scrolled down for about 2 hours straight, saving the images I liked. Not all of them had anything to do with forest spirits, but I also like to have some random images thrown in for originality and to draw on different colours and compositions, and in the end I had about 400 images in my inspiration folder.
I don’t usually gather so many, but I really wanted to push myself on this one, and spent the first week doodling concept sketches and gathering research – starting straight out in 3D can mean I don’t think about the composition enough.
In general, though, I draw on many different sources – as well as sites like Pixabay, I often look to old oil paintings by artists I admire – and for this entry, I also drew strongly on the incredible landscapes of the Hudson River school, specifically Frederic Edwin Church.
3. What software and plug-ins did you use to create this image?
All my concept sketches were made in Krita, which I like for its convenient access to changing brush size, type and colour, even though I generally just stuck to a small round brush, and sketched in black on white.
Then all the 3D work was made in Blender, with the main models separated into layers based on their place in the scene (Background, middleground and foreground). I also kept trees and the characters on their own layers to keep the working polycount down.
I used the Sapling addon to get the basic frame for the middleground and background trees, and the Manuel Bastioni addon to get the basic human shape.
For effects like the volcano smoke and the waterfall spray, I used a pretty old version of PaintShop Pro – it works and does what I need. In both of those cases I just used the default brush on a low opacity. I also used this program to combine the 3 layers which I rendered on Blender, and painted on the final colour grading, rather than going back into Blender.
4. Are there any particular techniques that you use often?
I really enjoy using Blender’s sculpting mode and used that on all of the main land features (I might have got slightly carried away though – the middleground model by itself reached 32,000,000 verticies…)
For the material, I used texture painting, and I use that all the time for landscapes. For an extra bit of randomness, I combined all the grunge maps I have into a node group, and mixed that with the painted texture, which I certainly intend to do more often.
Other than that, though, I try to vary my method, and it changes all the time.
5. Can you give us a short breakdown of your entry?
1) With all the research I collected, I made more sketches, the aim of which was to try and find an image which would contain everything I wanted without becoming noisy or cluttered. The shortlist of things to include (only about half of which I actually ended up with) was: A spirit with a creature, secondary creatures, evil creature(s), volcano, fire, trees, path with steps, water, fancy light patterns, a sad mood, rain/ash, eagles/birds, the sun, and other planets. Once I had that sketch (which is basically what the final render ended up as), I then drew out the background and foreground separately and annotated them with all the other elements I could include. This ended up at about 40 ‘things’. The last of my 41 preliminary sketches is shown in figure 1.
Fig. 2 – A late blockout render
Fig. 3 – a WIP showing the background in a late stage of completion
Fig. 4 – The painting which established the final scene’s appearance
2) Now I had a [reasonably] fixed subject and composition, I modelled a blockout in Blender, keeping it close to the sketch. I also added a human meta rig both for scale and to act as the character before I got round to modelling it, and made this as large as I could, since I have a tendancy to make characters quite small. In addition, I started to think about the scene’s lighting. (Figure 2)
3) Saving the file as a new .blend so I had the option of reconsidering the composition later, I then made the scene properly, starting with basic models with textures. Really, it was nothing more than adding a multiresolution modifier to the blockout models. The textures were initially comprised of the texture paintings.
After that, I then finished the background, and worked towards the camera, but didn’t make a lot of progress on the foreground at this stage. (Figure 3)
4) Once I’d inevitably become bogged down in the tedium of churning out countless small models and particle systems, I made a wip render and finalised the composition, excepting the characters, in 2D, by painting on the atmospheric effects. (Figure 4)
I also made the final lighting setup, after becoming bored with the old flat version. With this clear, I rendered the background, and painted on the various 2D additions, and did the same the next day with the middleground.
5) Now Sunday afternoon, I quickly sorted the foreground, adding in more particle systems and detail to the sculpting. Then I came to the real problem: the characters. I’d never properly decided on them, authough I had the human model. In the last few hours of making this I went through: stag, deer, swan (no, I don’t know why either) back to deer, eagle, coyote, buzzard and finally wolf. Annoyingly, especially as it was getting late, that still wasn’t working, and I had to add the pups and second adult. One of the reasons they look odd is that they’re actually the coyote models, and still have the coyote texture on the side facing away from the camera! (figure 5)
Fig. 5 – Camera view of all of the models
6. Any advice for people who want to learn 2d or 3d art?
Honestly, I think the best advice I can give is to tell people about the Weekly CG Challenge, and make them take part every time! It’s how I learnt, since before joining last November I was stuck churning out a mediocre animation, and wasn’t getting any better. But aside from that, there’s a lot to be said for studying as broadly as possible – look at all styles and mediums, especially traditional oil painting, because its been around for so much longer than CGI, and the whole area, in terms of image making, is more sophisticated than CG: at least for the moment. Composition is a topic I see many people struggle with, but if you get that right and can stick with it, chances are you’ll be motivated to put in the time and then the practical side of modelling and so on will come too. Always push for the next level!
7. Who are your favorite artists, traditional or digital, and can you explain why?
I don’t really have a favourite, partly because my interests in styles switch back and forth. I also tend to really like one work in someone’s portfolio, and feel a bit unaffected by the rest – I definately have favourite artworks, but that’s separate from the artist.
That said, I could certainly mention the contempory painter Geoff Hunt, who’s the past president of the Royal Sociey of Marine Arts in the UK, and his images of Nelsonic warships are incredible. I’ve spoken to him about his process, and the months of detailed research each picture takes is astonishing.
For historical painters, the Hudson River school member Albert Bierstadt stands out, along with the British painter JMW Turner, and the Renaisance painter Carravaggio. In all cases, they mix great realism with superb composition and lighting.
Thanks for the interview Gregory!