1. Creative upscalers - Part 1 Basics
2. Creative upscalers - Part 2 Tips and Tricks
3. Creative upscalers - Laundry List
4. Generative Fill - Part 1 Basics
5. Generative Fill - Part 2 Guided Generation

In the first major chapter we will delve into two tools, two different techniques. These are five lessons in the first chapter. They are quite technical, but also give you a lot of different tricks that simply make working with AI as faster and your life easier.

We create something from nothing, and it will be perfect in no way. But it doesn't mean we shouldn't have control over it. So, we show different aspects of when to finish, and how to start. What types of control we have in this process (e.g. generation through simple selection, and what that selection should be), what we can transform, what we can achieve, and how it affects our workflow.

When it comes to Creative Upscalers, we showcase all our tricks, going through everything technically. It might seem like using Upscale is just a click, but there are several hidden logics and functions we can take advantage of. One of the cooler lessons also shows when it works and when it doesn't.

Short and sweet lesson that aims to shift our mindset a bit. It emphasizes that not everything has to be perfect in 3D, and control isn't as crucial as you might think. We've always been giving up some control, but now we can do it completely honestly and consciously.

We’ll discuss three images. The foreground in the first one was done in 20 minutes, that building in something around an hour, and the facade of the church in 30 minutes. We show various tricks that can't be done in 3D because it would take too long, and can't be done in matte painting either because you’d need cutouts. But by combining both, you can take a nice shortcut.

7. Mediterranean - MATERIAL OVERVIEW
8. Mediterranean - LIGHTING OVERVIEW
9. Mediterranean - Tone-Mapping
10. Mediterranean - Deep storytelling
11. Mediterranean - 3d + Midjourney
12. Mediterranean - Post-pro - Firefly

The next chapter will be a break down of a Mediterranean interior scene.

The first lesson is Material Overview. We show in it how far you have to go with shading to get the most out of AI. Because if you do too little in 3D, AI will be practically useless. So, where is that sweet spot? How to manage albedo, glossiness and normal maps?

We break it down based on this scene and show some shortcuts you can take.

It’s less AI, and more 3D oriented lesson. Particularly 3D lighting-heavy, And it’s the last lesson based solely on 3D. This is the moment when we want to take care of the lighting in 3D because when the lighting isn't right, the AI starts hallucinating like crazy. And for all of you who’s been waiting for the interior masterclass, we'll extensively cover interior lighting, step by step, and explain what to do to keep control.

Another lesson will be a treat for all those enthusiasts who love color and enjoy geeking out on the technical aspects of it. Not gonna lie. This is a very technical lesson, but if you're not comfortable with the frame buffer, it will be very useful to you.

Toward the end of the chapter we're returning to integrating AI and 3D and make our first approach to embedding a character in a full scene. The workflow is very, very funny, and our 3D output is straight up ugly, but we'll make it pretty in post-production. This whole workflow is meant to inspire you to approach post-production with the anything-is-possible mindset.

13. Penthouse - Deep Storytelling 2
14. Penthouse - Midjourney - Part 1 Basics
15. Penthouse - Midjourney - Part 2
16. Penthouse - Midjourney - Part 3
17. Penthouse - Midjourney - Part 4 Prompt Commentary
18. Penthouse - 2D do 3D - Tripo3d
19. Penthouse - LIGHTING OVERVIEW
20. Penthouse - post-production

Okay, another big chapter, and another way to dive into deep storytelling. Here we discuss a penthouse scene, it's a nighttime scenario, so as you might expect, we will be talking about lighting. The approach here will be different though, more advanced, but also very fresh.

There are four lessons on character generation in Midjourney. We will prompt until we get a character with the right angle, the right style, and the right activity. First lesson is AI-heavy. We show you what worked and what was the thought process behind it. If prompting feels a bit intimidating to you, I feel this is a good place to get started with Midjourney.

Lesson 7 of this chapter covers nighttime lighting in interiors. It will involve a different logic, a different process, focusing on different aspects, and different color management than daytime lighting. We'll go through it step by step with additional commentary and analysis in our proven CommonPoint style.

And in the final lesson we put it all together. This is a great example of real-life post-production involving AI and 3D. We encourage you to use AI to disrupt your old 3D workflow, from modeling, shading, lighting, to post-production.

This is just a single, short but quite hefty lesson focused on Parisian interior scene. We’ll walk you through the scene, starting with the lighting, the critical elements, how the scene is constructed, and finishing with the compositional intent.

We'll revisit what we did in Midjourney, and talk a bit more about prompting: what worked, what the process was like. We use similar tools as before, but show you even more examples. It gives you that confidence that maybe if I approach it this way, it will work out too.

Another short lesson, another interior, another story from start to finish. All those elements were created by AI. These are furniture pieces generated by Midjourney, and than turned into full 3D geometry.

It's a handy technique for when clients hand us some terrible models, sometimes from antique shops, and there’s no way to find the models anywhere.

Of course, these models aren't perfect, but again, we encourage the idea that you don't need hyper-realistic output and you don't need to get all the details at the 3D stage. With AI you can do that in post-production. We go through many different examples, and various ways to recover those details in post-production.

23. Exterior - SCENE OVERVIEW - Arch/Landscape
24. Exterior - SCENE OVERVIEW - Entourage
25. Exterior - LIGHTING BREAKDOWN
26. Exterior - Post-production

We talk about building a scene and what needs to be taken care of. Now, we can actually have great-looking people in our scenes, (well, we can - after upscaling) so it's worth including them in. But if you haven't been a post-production artist, you might not know how to set up entourage directly in a 3D scene. So we’ll walk you through the process.

We talk about assets, how to select them and why, how to place and arrange people and make sure they are diverse enough. Because the design decisions we make, will be embedded in the image. All AI can give us is production value.

We also talk a bit about the lighting, and do a small recap about maintaining control over lighting in exteriors, just as we did in interiors. We discuss some cool aspects, and finish it of with post-production. We refine the details of our imperfect 3D models. We’ll use many techniques, and I think you’ll be interesed to see the level of care required to create really great images.

This is a crazy lesson, where beside all the lighting, framebuffer, and so on, we also heavily cover post-production. We show you again how to push AI to the absolute limit, how to manage detail, how to force those hallucinations, and a lot of post-production generation enhancement. Those techniques are very cool, very organic, and very effective.

28. Aerial Making Of
29. Aerial Advanced Style Transfer 1/3
30. Aerial Advanced Style Transfer 2/3
31. Aerial Advanced Style Transfer 3/3

In this chapter we create an aerial, and all four lessons are about changing the lighting scenario with zero rendering. We change sunset to day and sunset to night.

It’s quite an experimental workflow, but it can really help you realize 3D isn't the most important thing. Sometimes you can just pull a rabbit out of a hat, and it comes in handy.

The first lesson is sort of a making-of, where we describe not only all the technical aspects we used—e.g. a lot of post-production —but also the “whys”. Like why did we shift that mountain 5 pixels to the left? Why that palm tree is that particular size? And so on. It's not just a technical approach but a creative one to understand how the geometry of the image works.

And we’ll talk about all of this in those very, very cool four lessons of the chapter.

In this lesson we discuss the story of the Polar Expedition image. We show you how we create great storytelling assets and draw various inspirations.

This image is also sort of a test of what we need to get close to the inspirations generated by Midjourney as much as possible. What limits us? And what the sketching process looks like? Because that’s what this image generally is - a sketch. It's really fun process that shows you how you can create things, and slowly enter the world of concept art.

33. Silk Road - Collaborative mindset
34. Silk Road - Research
35. Silk Road - Research Midjourney
36. Silk Road - Research Leondardo / Magnific
37. Silk Road - Asset Creation
38. Silk Road - 3D Scene/Rendering
39. Silk Road - Post Production
40. Silk Road - Feedback Loop

The last chapter is the Silk Road and the Kathmandu Image. This is the most detailed project of the entire masterclass, with eight lessons covering many different topics.

At the beginning, we talk about the backstory of this image, why it was created, how to approach such concept art or matte painting images, and how to create them.

Absolutely everything is broken down into its components. From research, that goes waaay beyond a ready-made mood board. We literally start from how to educate ourselves to understand this reality, how to find key frames, and what the organic process of searching for color frames looks like in Midjourney, Leonardo, and Magnific. An enormous amount of research. One of my favorite lessons.

And then, of course, knowing the rules of this world, we need to create assets. Tons of assets. 2D and 3D. In different positions. With different lighting, etc. There's a whole lesson about the process and creating this frame from scratch. A timelapse showing what we set up and why? What the compositional reference is?

It’s obviously a very packed lesson but even after post-production, we’re still not done. At the very end there's a separate lesson where we apply feedback. Based on this image, we have two rounds of revisions, which Bartek was very excited about. You can watch how it genuinely looks in practice. It's great to see that in real-life situations you can manage feedback with this workflow, and that 3D isn't the most important thing anymore.

In the first two update lessons, we replace people using Nanobanana. We get that nice realism faster, but also run into some issues.
So we discuss where Nanobanana works well, where it doesn’t, and how to deal with all of it. We also build storytelling directly in the image, starting both from placeholders, and from an empty scene. And finally, we show you how to clean up after scale and aspect ratio problems you’ll face when working with Nanobanana.

We get back to generating cutouts. In the training, we use Midjourney to curate characters. Though, getting the right angle with it can be tricky. Luckily, ChatGPT and Nanobanana let us create cutouts much faster and adjust the view with better precision. Sometimes we even let the AI place characters directly into the scene. Spoiler alert: it doesn’t always work, but we show you different tricks that help make that happen.

In Lesson 20b, we test if we can create the same storytelling we did for our “staircase” image. But even faster than before. Without Midjourney, generating models, or creating custom cutouts. Just one image to start with. Can it be done? Yes, but by no means is it a one-click solution. We’ll guide you through a few problems we run into along the way. Plus show you extra tricks to reach the results you want. And also take this opportunity to discuss the pros and cons of working with Nanobanana.

In this part, we explore more advanced techniques for those using AI in interior design. Starting with generating textures, creating normals and roughness maps. We test if we can get usable, flat textures from low-res photos of wrinkled fabric, clients often send us. We also show how to transfer details, when our models lack quality. Just like we did in previous interior lessons. You’ll learn how to “hallucinate” beautiful carpets and realistic, soft cushions. And in the third lesson, as a fun experiment, we do AI interior staging. We feed AI an empty interior and a furniture collage. And we let it design the space. As always, we discuss both pros and cons, and give you tips on how to deal with problems along the way.

In this chapter, we jump to all the latest techniques you can use for exterior scenes. We share tricks from our everyday commercial projects: from useful shortcuts, to completely transforming the entourage. You’ll learn how to replace people, change their ethnicity, age or, for example, turn them into college students. All of it, taking advantage of the latest AI tools. In another experimental lesson, we hallucinate interiors, generating them “for free” inside existing exterior shots. We also test creating different lighting scenarios or changing the camera to show you the range of the new creative possibilities ahead.

In this final update lesson, we go back to our Silk Road image and do post-production all over again. This time, using the latest tools. We’ll compare one-to-one the results from a year ago to those we can do today. And comment which tools work better and which not-so-much. We hallucinate clouds, or try to recover details in the monks’ robes “for free”. And even though not everything works out, we get some unexpected results that weren’t really possible with previous AI tools. Like a completely new winter version of the image. So it’s definitely fresh, new and exciting. And we really hope you’ll find it all useful.