$209.06
$278
You save $68.94
Bundle discount Each additional course part in your cart adds 3% off all courses, up to 15% off.
Give a gift
Give a gift card for art students to use on anything in the Proko store.
Or gift this course:
I Teach Creatives
LESSON NOTES
Check out the premium course for additional lessons, demos, assignments and critiques!
Boxes are the basis of complex perspective. Control three dimensions and you control depth on paper. We can see complex objects as a series of cylinders. Cylinders have complex curves that we best understand by reducing them to boxes and straight lines.
The Simplest 3D Form
Let us see how simply we can simplify form. A shoebox is proportionally complicated with different heights, widths, and depths. A sphere has a single curved plane. Curves are complex. Triangles are difficult to draw in perspective.
The most conveniently simple three-dimensional form is the cube. Boxes are made of right angles. Every line is at a right angle to each other line that it touches. The cube is even simpler because it is square all around. Every plane is a one-by-one square.
The Challenge of the Cube
Simple does not mean easy. Each plane is exactly the same size and shape. This is a hard criteria to meet. Squares tipped into space do not appear square due to foreshortening.
Mastering foreshortened squares gives you control over proportions in space. They solve the problem of field of vision. They help establish how dramatic you want depth to look. Constructing the most basic form in the world, the sphere, comes clearly by seeing the cube as its parent.
Methods for Drawing Cubes
Here is how to master the cube. Study it. Draw from it over and over. Imprint it into your brain.
Shapes First
Rough in planes lightly and loosely. Compare the shapes and eyeball the angles. Do not fuss over one plane so long that you forget it is only one part of the cube. See the whole cube as a thing. Let the line directions come from that view of the whole cube. This improves your perception of how squares turn into new shapes.
Get in the habit of fearlessly and freely beginning a sketch. Gradually hone your precision. This avoids a self-conscious and stifling approach. Judge your drawing lightly instead of harshly. See this as a process rather than a product. Delaying your commitments by gently seeking line directions trains you to see the correct lines in advance.
Line Systems
Start with the X, Y, and Z axes. Put down the three crucial lines that give you a front corner. That helps you find other corners at varying angles. Right angles now become acute or obtuse angles. This method relies on your knowledge of where lines go in space.
Construct Within a Circle
Cubes do not stray far from the ratio of circles. Start with a circular blob. Find your axes and seek convergence.
Receding Convergence
Boxes have parallel lines. When parallel lines go away from our face, they converge at a vanishing point. This is receding convergence. Lines aim away to unite like they should.
One of the most common errors in freehand perspective is drawing lines that do not look like they go away. We call this receding divergence. Lines spread apart instead of coming together. This is bad. It takes conscious attention to fix this until it becomes a habit.
Check out the premium course for additional lessons, demos, assignments and critiques!
DOWNLOADS
ASSIGNMENTS![]()
![Daniela]()
![]()
![Dooby]()
![]()
![Dooby]()
![]()
![Carlos Javier Roo Soto]()
![]()
![Marshall Vandruff]()
![]()
![shaswat]()
![]()
![Carlos Pérez]()
![]()
![Sandra Salem]()
![]()
![M C]()
![]()
![Michael Giff]()
![]()
![Max Long]()
![]()
![R081n]()
![]()
![@rupertdddd]()
![]()
![Max Long]()
![]()
![Ash]()
![]()
![Espy]()
![]()
![Sandy]()
![]()
![Lin]()
![]()
![Michael Giff]()
![]()
![Wenhan Lee]()
![]()
![Sajal Khare]()
![]()
![Chuck Ludwig Reina]()
![]()
![Ishaan Kumar]()
![]()
![Chuck Ludwig Reina]()
![]()
![Carlos Javier Roo Soto]()
![]()
![Marshall Vandruff]()
![]()
![Max Long]()
![]()
![Daniel Lucas Nizari]()
![]()
![Lin]()
Draw 100 rough cubes freehand. Add little arrows to show where each set of parallel lines is converging in space. Focus on making the lines feel like they recede toward the same point, not diverge apart.
- Draw quickly and loosely
- Don’t use a ruler
- Think about line direction and convergence
- You can use a real cube, make one, or use the Zolly app cube model
Deadline - submit by June 07, 2026 for a chance to be in the critique video!
Shout out to this one comment explaining how cubes and spheres relate to one another like 3 lessons in. The sphere mention reminded me of this and I'm so glad I saved it back then https://www.proko.com/s/N15G
Also want to share this little animation I did. This is what the Zolly app basically does, but I feel like recreating it frame by frame really helped heighten my awareness of the Z Axis going AWAY in space
In that one "rapid fire" page I just tried to quickly draw the box in ~5 seconds and then later cleaned up the lines a little. I know all of these boxes are meant to be quick and loose, but I couldn't help myself and wanted to take it a bit slower, meticulously thinking about were the vanishing points are headed.
...Aaaand 100!!! *feew I'm not gonna lie I kind of feel defeated by the end. Almost like I failed my mission to kill the receding divergence. Even though I tried everything Marshall show in the lessons and even give them names to the shapes of each plane. I guess I'll just have to do 100 more.
Any feedback on what I could have done better?
•
10h
Carlos - good job!
For this project, I see nothing you could have done better with your product.
With your process, however, it would be better, if possible, to find a way to feel victorious rather than defeated.
You could have fooled me. You could have claimed "I really got a kick out of trying to craft each cube with converging lines and nameable shapes" and I would have believed you, because these look like they came from that.
My recommendation for you now is to set cubes aside until our next lesson group, where we'll refine angles. You deserve to spend some time goofing around, sketching whatever pleases you, with a mild eye toward it's line systems and how things "plane up," but only a mild eye.
In a couple weeks, we'll get back to cubes with another challenge. Going back and forth between conscientious practice and playing around will help you balance the process.
A hundred! I had some issues with receding divergence in the beginning, but by the end I got the hang of it.
Cube control , at the beginning I was too perfectionist maybe I used silly for like the first 40 , then like 20 from memory and at the end I used some references from Kim yin gi , I saw how he put them close to each other and playing with dimensions closeness and so on
I fell short by 30 boxes. I realized after placing the ellipses inside the boxes I was not getting accurate cubes. I wonder how artists can predict what a cube would look like in 3d space. Also, how can I make sure that an ellipse is correct without relying on intuition. Looking forward to receive the answer to these questions and get back to work.
Keep an eye for the second submission to this homework, I will be more playful with the remaining 30 cubes!
I don't think you're as far off as you think. Like Max has said the arrows do not go both ways. If you're having trouble knowing which way they shoot off I would recommend starting with the front corner. I traced over some of the lines of your reference.
Hope that helps.
The one thing that I see that could be considered wrong is that the arrows are pointing in both directions of each line. It is my understanding that the arrows are meant to show the direction of the vanishing point on the horizon line. If your reference cubes are from the Zolly app, you can spin the cube horizontally left or right and the vanishing point will appear somewhere on the horizon line. For the vertical lines, if the cube is above the horizon line, the vanishing point will be above the horizon line; if the cube is below the horizon line, the vanishing point will be below the horizon line. Hopefully, this makes sense?
Hi! I think it is looking pretty good, you just need to keep practicing, the main thing I see is the line quality, you should keep practicing ghosting the lines and doing the warmup exercises.
Also, Have you tried drawabox?
Here are my cubes; I tried a systematic curvilinear perspective approach after two pages of random cubes. Ran out of steam but If I had completed I would have 177 cubes on this page!
As Ash said, that third image is awesome! That’s an awesome take on Peter Han’s Head lesson, as most head drawing courses imply that a head can be represented by a cube. I may have to try this one myself. The other two images are great also. Thank you for posting this.
•
1d
That 3rd image is awesome!
Here are 101 cubes! Directions shown via the shaded front planes.
Proportion consistency seems to be the biggest problem I had while doing this. I think I often tend to exaggerate the convergence too much, making the boxes appear to have more of a "narrow lens" look, and thus it can make the boxes seem more like rectangles at times.
I was slowly getting a bit more used to them towards the last couple of boxes at the bottom, though! (After some redlines/self critique) Not perfect, but we're aiming for progress, not perfection!
Marshall, I am having some trouble finding the top plane. I have been starting with the Y lines, however as I build everything out and draw the through lines, it becomes apparent that the perspective is not correct. All of these boxes are from the Zolly reference and almost all of the top planes end up too wide. I also have difficulty making it feel like all the sides are even and feeling like a cube rather than a rectangle.
This is the hardest possible part of a box for me too. Try with an ellipse which confines the space and margin of error and thus makes it a bit more manageable. They’re all trapezoids within that ellipse. It’s going to be really hard this way too but with a bit of time and observation we can get closer. My process now is:
1. Decide how open the ellipse is aka how much top plane we are seeing
2. If we are seeing a lot of top plane think about how much the rest of the cube is visible. It’s all connected
3. Decide how much side is visible and mark it with a vertical
4. Finally draw the trapezoid. It doesn’t have to be perfect. These below are not. It just confines you to an area/HL instead of overshooting everywhere
I used to try draw it directly but it helps to make these decisions first, it becomes more manageable for me because it removes the overwhelming number of factors involved in drawing that top plane into bite size steps.
You can also make a cardboard circle with a square inside and look at it from various angles to learn what it looks like when it becomes a trapezoid .
lastly, this stuff involves a lot of mistakes and I found I can’t remember specific angles or my many many corrections. But what I try to remember is paths, aka how to get the effect I intend when something slides off. For example red (1) is what I’d do to a plane, where I give it too much like you, and black (2) is the correction. It’s achieved simply by “sliding” the far edges by shifting one axis. So now when things look wonky I do this and often it fixes the problem. I found learning how to fix stuff when it’s broken (observing what shifting an axis does to the look of a plane) is more important for me than learning to make the cube great on the first try, because it teaches control and with control you can get to a result eventually. I don’t have to remember a billion things about what planes look like in various angles, or what proportions a cube has, if I can just shift stuff until it feels right. I hope that makes sense and helps a little.
I also struggle mightily with this. Even if by some miracle I get the top plane to squish down proper the line systems will not longer match up.
What I've learned is that drawing a cube/box by itself is easy, but drawing one with certain constraints can be substantially more difficult. For example, drawing 3 cubes in a straight line is more challenging than drawing 3 cubes individually with random angles.
This assignment was fun for the most part, except that one time where I had to redraw a cube 9 times just so that it looks right.
Feel free to give constructive criticism :).
What does marshal mean by roughly drawing the boxes and slowly making them more precise
•
5d
Not to put words in his mouth, but I think what he means is to lightly sketch in the box, and then with those searching lines, find the corners and lines and more heavily draw those in.
The big lesson ('perspective'?) that I gained from this exercise was exaggeration of angle. The first half or so of my submission seems more unconvincing than the latter part because I tried to draw what I saw. But then I remembered Rembert and his recounting of his teacher's words about exaggeration and realised that visual accuracy is more effectively represented on paper with that extra push of making angles more acute or obtuse, and/or drawing certain lines shorter or longer than your instinct tells you to.
•
5d
That's a great point! It's also interesting how these little things we hear might not make a lot of sense when we first hear them, and then we get into a drawing problem and BANG it all comes together.
(Also nice boxes!)
Few!! The first 50 cubes and hopefully I'll be done with the rest by Wednesday. In the mean time I want ask for advice about how to kill the receding convergence more effectively, each cube took me from 3-5 minutes. I don't know if that's too quick or too slow enough for been roughs, and as you can see I have some troubles with keeping clean some of these. I think is because I was too focus I getting it right with me reference and, trying to remember what corners are acute and obtuse when they should be.
But anyway I'll appreciate some advice to kill some of these errors before they become bad habits, or worse if they already are bad habits.
In that case, any plan to deal with them is welcome.
•
10h
No bad habits here. They are clean enough. Some proportional struggles, but that will be our focus in good time.
3-5 minutes is not too short a time. It's about right.
This is exercise to get you seeing. All good.
100 freehand rough cubes done. Every one of them were done from the Zolly app. The convergence of the lines were something else. Getting the angles correct was essentially nonexistent. I don’t think any of them were perfect. I changed the routine up due to monotony for one page with the XYZ axis, and two other pages with the cross from the primitives section of the Zolly app. Also, I tried to convey the line weight as shown on the Zolly app. I probably need to do about 900 more rough cubes in perspective! 😧 I kept having memories of that time I had to write a thousand times about not talking in class without permission. But, I did learn from that experience, just as I have from doing this assignment. As always, critiques are welcome. Cheers!
I suspect that I keep changing my horizon line for each set of free lines. I noticed that the smaller side of the cube convergence and also if you know a specific few sides, you should be able to mirror them for other positions.
overal sometimes I feel my starting x y and z line are either in the wrong degrees apart or too long or too short
Hey, that’s the exact issue I’ve been battling for a year. I can converge okay but the two families will go to two different horizons levels no matter what (when trying to make the box parallel to the ground), and you’re the first person I’ve met to name this issue so far. 😅 It does come from an improper expression of the xyz perpendicularity, I think? Because it lifts up a box corner almost like it’s sloped. That being said my brain can’t hold on to the HL no matter how hard I try or how many exercises I do. BUT I discovered in this assignment that learning how xyz (90 degrees) look in space and doing it intuitively is helping me far more than trying to figure out the HL. It’s an alternative route to arriving in the same place I think. If your brain is anything like mine, maybe focusing on the axes could be the way of this dilemma. I feel space like a 3D scanner in those sci-fi movies where you enter a cylinder and it scans you from top to bottom with an ellipse of light which changes openness based on how far it is from the HL. And each “level” looks a certain way. So I use this method to feel space. on a box if you draw an ellipse the points will be on that ellipse. If this resonates maybe this below will help (just the outer ellipse along with marshall’s sphere is enough imo) to keep the two families “level” - it is what made me be able to do it better after a year of struggle: https://youtu.be/wJ_k4aCknZYsi=Y0ygsK8riCYW_8pNv
