Thirty minutes is enough for one question. I spend most of it drawing by eye, and bring out the tools only once I’ve committed to an answer.
For the last five minutes, I hide the source and draw the same relationship again from memory, or on a different subject. That sketch is often ugly, and it’s the useful one: it shows what stayed with me once the picture was gone.
Polish hides what went wrong. If the proportions drift but the value grouping works, a finished copy won’t tell you which of the two you got right. With one question, you know what the session taught you, even when the answer is “I misread it.”
Choose one question
Pick an image where one problem is easy to see. Strong lighting separates the value groups, and a limited palette shows how much room a small accent takes in a quiet field. For proportion, I want a pose whose long axes still read at thumbnail size.
If you need images to choose from, the free art reference guides collect museum collections and gesture drawing sites with the reuse terms already checked.
Before I draw, I write the question in a note on the board. It might be:
- Which three value groups make the light read?
- How much of the frame does the warm accent take up?
- Which large shapes lead to the focal point?
- Where do the hardest edges sit?
- How long is the torso compared with the head?
Once the question is written down, everything else can be messy. In a value study the contour can wobble, and in a proportion study the shading can stay unfinished. The rest of the reference can wait for another session.
Make a small study without aids
Dorian Iten’s value-study assignment at Proko gives each attempt 15 to 30 minutes. The time limit forces tonal relationships to come before detail. Use the same limit here: keep the reference in view and the ToolBox panel closed, and work small enough that you finish.
For a value thumbnail, three to five flat tones are enough. Put down the big light and dark groups first, then decide whether the image needs one middle value or several. If three tones explain it, stop at three.
A color study works the same way. Pick a handful of swatches and guess how much of the frame each one covers before you block anything in.
A composition study can fit in a frame with four or five shapes. For a figure or an object, I choose a unit to compare against, like the head or the width of the ribcage, and build the subject from simple forms. It’s harder than following a contour, because I have to account for depth.
Stop as soon as you can answer your question. Save the polish for another drawing.
Check the attempt with the source
Don’t fix the first attempt, even when the mistake looks obvious now. In ArtDeck, Snapshot to Board puts each check on the board as its own image, so it sits next to your drawing and your first judgment stays visible.
Value
Switch the source to grayscale and compare the biggest masses before the small patches. A two-value Notan is blunter: shapes either join or they don’t. As you move the threshold, watch for a dark you split in two that the source keeps as one shape.
I check range with grayscale and grouping with the Notan. They catch different mistakes, and you need to know which one you made to choose what to practice next.
Color
Put your guessed swatches on the board before you extract a palette in the Color lens. Otherwise the tool answers the interesting question for you. Compare value and temperature first; the exact hue can wait until you know how much room each color takes.
In the lake photograph above, the two oranges each cover four percent of the image, and the pale blue-grey covers 38 percent. A study built around “autumn orange” can easily give the accent too much room.
Write the corrected relationship down as an instruction you can reuse. “A cool field with two small warm accents” will help you more than a row of hex values.
Composition and edges
Put the shape map next to the source and look at the crop and the quiet areas. They often tell you more than a composition grid does. I keep the grid for an alignment I can’t find by eye, not as a stamp of approval.
Then squint at the source, or blur it with Squint in the Read lens until the small detail disappears. Which edges stay firm? Where does one shape melt into the next? If your study has the same edge around everything, pick one focal edge and soften several others.
Proportion
Pick one length as the unit and commit to an estimate, like “the forearm is about three quarters of the upper arm.” Write it down before you measure. The ratio then shows how far your estimate drifted: 0.75 on paper, 0.62 in the source.
The measurement tells you what to try next. Next to the untouched study, add two notes for the next drawing: one judgment that held up, and your biggest miss.
Close the reference and draw it again
Now hide the image for real, with no little thumbnail left in a corner. Give yourself five minutes to draw the study again from memory, and forget about finish.
When you bring the source back, compare the big decisions first. If the main value split or the proportion is gone, look at that part once more and try again.
Proko’s visual-memory exercises use the same loop: observe, hide, draw, check. Harder versions turn the subject or change its proportions.
Memory drops detail fast. What stays is what you understood or found distinctive, and what’s missing shows where your hand followed the image without really understanding it. That’s where the next attempt starts.
Use the lesson in a different image
Now use the idea on a different subject. Carry over the decision you studied, and leave the layout of the original image behind.
A new landscape can reuse the same three-value relationship, or a costume can borrow the ratio between a warm accent and its neutral field. A figure can keep the gesture with a different body type. The one sharp transition from an edge study can go to a completely different focal point.
This drawing can be crude. If the idea falls apart as soon as the subject changes, go back to the study and write the relationship down more precisely.
Keep the first attempt
Keep the reference, the first attempt, the check, and the memory or transfer drawing together. If you only save the corrected version, you lose the judgment that needed work. The awkward first attempt is also the one that shows, a month later, how much your eye has changed.
After several sessions, the same misses start to show up. My compressed darks might call for a week of value thumbnails, and drifting long axes for more estimates followed by ratio checks. Your next exercise comes from that pattern instead of a generic practice list.
Then make something you care about and carry one relationship from the study into it.
The study tools page covers every lens, and an ArtDeck reference board keeps a project’s sources and studies together. If the reference moves, the frame-by-frame workflow shows how to find the frame worth studying.
Common questions
How long should one study take?
Thirty minutes is enough for one question. Dorian Iten’s value-study assignment at Proko gives each attempt 15 to 30 minutes, and the short limit is what puts tonal relationships ahead of detail. Stop as soon as you can answer the question you wrote next to the study, because a polished copy hides which judgment failed.
Is this just tracing or copying with extra steps?
No, and the order is what keeps it from becoming that. You draw by eye first and commit to an answer, then use a tool to check the one thing you guessed at. If you extract a palette before writing down your own guess, the tool answers the interesting question for you, and that’s how a study turns into a smaller copy.
Do I need to be good at drawing to get something out of this?
No. The checks are most useful while your eye is still learning. A two-value Notan or a palette’s percentages tell you something true about the image, whether or not you draw it well afterwards.
You can also run a lens over a picture just because you want to know why it works.
Why keep the first attempt if it is wrong?
Because the corrected version hides the judgment that needed work. With the reference, the first attempt, the check, and the redraw kept together, the same misses start to stand out after several sessions, and they tell you what to practice next.