TALAMANA · THE AI LITERACY MAP FOR ARCHITECTURE AND DESIGN · Judgment · AGE 17—20 · POSITIONAL · HELD
The batch is a contact sheet
A single frame shows what the machine can do, and the set shows its tendency.
The dilemma
Sixteen images come back from one brief. The third one is strong. Your eye has already stopped on it, and the other fifteen are turning into background. Do you open the strong one now, or do you make yourself read the whole sheet before you look closely at any single frame?
The choices
Open the strong frame first and judge the rest against it. Or scan the whole set at one small size, frame beside frame, and only mark the few worth enlarging once the scan is done.
The consequence
Open the strong frame first and it becomes the anchor. Every other image is now "less like number three," and whatever the set as a whole was telling you is lost. Scan first and you see what the batch agrees on, where it disagrees, and which frames only mean something because of the frames beside them. Enlarging comes after, and it is quick, because the scan has already done the work.
The case
A second-year student pins up one beautiful image of a stepped ghat-side pavilion. The tutor asks to see the rest of the batch. Eleven of the twelve put the pavilion on the water's edge. The brief never asked for that. The one image the student chose was the only one that did not. That is interesting, and the student had not noticed it, because they never looked at the twelve together.
Our position
The core skill in reading generated work is fast comparison across the whole set, before any single image pulls you in. Treat the batch as a contact sheet. Until the sheet has been read, the one strong frame is a distraction.
Why we hold it
The batch is the only evidence you have of what your brief actually asked for. A single frame shows you what the machine can do. The set shows you what it tends to do, and the tendency is the useful information. Photographers learnt this with a loupe over thirty-six frames. We have taken the habit and apply it to sixteen.
The strongest objection
Reading in batches creates a bias of its own. At thumbnail size the most striking frame dominates the comparison and the subtle frame disappears, so the sheet trains you to select the loudest image, not to judge the set. The frames are not independent witnesses either: sixteen draws from one prompt share one set of habits, so "what the batch agrees on" may be the model's habit rather than a finding about your brief. And a contact sheet shows you only what was generated. It cannot show you what the brief should have produced and did not. Comparison trains selection. It does not train judgment of what is absent.
What would make us revise it
Workshop logs showing that scan-first students pick the loudest frame as often as frame-first students, and miss the absent idea as often. The scan would then be a ritual rather than a method, and the card would be rewritten around the harder question — what is missing from this set — with the sheet as one tool for asking it, not the first move. We run both methods in workshops and read the log.
Try it
Scan a set of twelve to twenty images at thumbnail size. Mark three worth enlarging, each with a one-line reason. Then point to one thing a frame shows only when set next to its neighbours, and not alone.
Take it to crit
Watch how the student moves through the batch. Do they read across the whole sheet before enlarging anything, or do they fall for the first strong frame and treat the rest as worse versions of it? Ask what the set agreed on. If they cannot say, they never read it.
How it works
A contact sheet works because the frames are small and side by side. Small size stops the detail from seducing you. Side by side lets the pattern show. Look at the set and ask three things: what do most frames do, where do they disagree, and which frame stands out. It is the same reading you learnt on a chart of numbers, now on a grid of images. Only then pick the three worth opening, with a reason for each. This card is about seeing the set. The next one, DIAGNOSIS, is about reading what the set says about your brief.
What this idea builds on
What this idea opens up
Sources
- dreaming-engines-and-grounded-generators
- dreaming-vs-grounded
- HCM.A
Open this idea on the map · The complete map · Logika · RBDS AI Lab, India · revised every edition.