A renderer you can defend — Parallax Sketchbook

SKETCHBOOK / 12 The project table · 2-week chapter

Page 1 of 4: Choose one finishable promise

Scope is an engineering decision

Choose one finishable promise

Choose one project: a one-bounce mirror study, a tiny raster postcard, a perspective checker comparison, or an original procedural shader. Write a sentence describing its input, visible behavior, and limit. Keep other ideas in a later list rather than quietly turning this project into a new engine.

The supplied code is a reference and scaffold. Your portfolio contribution is the change you designed, the evidence you collected, and the explanation you can give without the lesson open. Do not present an unchanged demo or an agent-generated implementation as proof of your own mastery.

Define the stop line now: one reproducible command or browser workflow, one representative scene, one tested edge case, and one known limitation. A small finished study can communicate more than an ambitious repository with no runnable result.

TRY IT IN YOUR STUDIO

Write your project promise and a three-item finish list. Name two features you are explicitly leaving for later.

From the page to your own work

Your studio assignment.

Finish one small rendering project whose decisions you can explain without the course open.

A finish line you can see

Your own bounded renderer study with source, one reproducible scene, a short demo, tests, a debugging story, and a measured result.

Pick one extension and explicitly defer the rest.
Produce an original visual or behavior plus a design explanation; unchanged scaffold is not a portfolio project.
Record the exact build/run steps and one known limitation.
Answer a geometry/ownership question and a debugging/performance question using your code as evidence.

Keep the evidence in your weekly notebook ↗

Explain it like an engineer

Try answering aloud before opening the review notes. Use your own implementation as evidence.

Tell me about a rendering bug you diagnosed.

Describe the observed result and a small reproducible input.
Explain the hypothesis and evidence that isolated the failing stage.
Show the fix, regression test, and remaining limitation.

What did you optimize, and how do you know it helped?

State the workload, build mode, machine, and repeated measurement method.
Separate CPU time, allocations, and GPU timing when relevant.
Report the tradeoff and avoid claiming an unmeasured improvement.

These prompts support practice; they are not an automatic assessment of your work.

A tip for this exercise

Choose the extension you can explain most clearly, then freeze its feature list. Spend the final sessions on a reproducible result, one useful test, and the story behind your own decisions.

Hints, when you need them.

Try a first step, then reveal as much help as you need. Each hint gets a little more specific.

1Hint 1 · Find a starting point

Write one sentence naming the input, visible behavior, and limit of your study. Mark which parts came from the course scaffold.

2Hint 2 · Choose an approach

Reduce one real bug to a small case with a known answer. Keep its regression, and measure one fixed workload before and after a single proposed change.

3Hint 3 · Check your reasoning

Rehearse a five-minute review: reproduce the result, trace one pixel or sample, show your regression, and state a limitation. Include build mode and repeated measurements even when the proposed optimization did not help.

A small creative variation ↗

Make a second scene using the same feature and pipeline. Improve presentation before adding a second major subsystem.

Optional. Save it for later if this chapter already feels full.

Coming back after a break? ↩

Write a three-item finish list: one broken thing, one missing explanation, one reproducible command. Choose only the smallest item today.

Books & principles behind this study +

Original Parallax explanations and assignments, with these works for further reading.

YOUR MARGIN NOTES

Leave yourself a good question.

What changed your mind? What would you try next?

Notes for A renderer you can defend

Notes are private to this browser and saved as you write.

A CHECKPOINT, NOT A SCORE

Can you explain the idea and try the exercise yourself?

Completion records your own assessment. It does not compile or test your C++.

Companion source 4 files