Screenshot Quarterly
Games, framed
Every number in this publication comes from one of the three procedures below. All three are gameable and all three have known weaknesses.
Screenshot Quarterly
Games, framed
Every number in this publication comes from one of the three procedures below. All three are gameable and all three have known weaknesses.
Three procedures produce every figure here: a frame log kept during play, a review of that log exactly one week later, and two ten-point scores awarded afterwards with the log open.
Every frame taken during a review period is recorded at the moment of capture with four things: the focal length if the game reports one, the subject type, the simulated time of day, and the location. Nothing is deleted during play, including obvious failures, because the denominator is half the measurement.
Frames taken accidentally are logged and marked. There were nine of them this issue and all nine were discarded, which is the expected result and not an interesting one.
Seven days after the last session, the whole log is reviewed in one sitting at a consistent size on a consistent display. Each frame is kept or discarded on a single question: would this still be worth looking at if it came from a game I had never played?
That phrasing does specific work. It excludes frames that are interesting because of what was happening — a rare event, a difficult fight, a joke — and keeps only frames that survive as images. Several of the most memorable moments in every game reviewed here produced discarded frames.
The ratio is kept frames over frames taken, expressed as a percentage. It is the only figure in this publication that resists argument, and it is also the one most easily distorted, which the next section covers.
Experimentation inflates the denominator. A game with a deep photo mode invites you to try the same composition at four focal lengths and three apertures, and eleven of those twelve attempts enter the log as failures. A game with no camera at all produces only deliberate frames.
This is why Glasshouse Rose, with 214 frames, appears to outperform Causeway, with 901, by eightfold. Some of that gap is real and some of it is an artefact of how frames get taken in each game, and the reviews now say so explicitly rather than reporting the ratio unqualified.
Learning also inflates it. The first hours in any game produce poor frames, and a game reviewed at thirty-eight hours carries more of that dead weight than one reviewed at fourteen. No correction is applied, because any correction would require deciding when learning ended.
Awarded against a fixed list of twelve controls: focal length, aperture, focus placement, shutter, sensitivity, roll, tilt, height, time of day, weather, interface hiding, and post-processing. A control counts only if it changes the image rather than decorating it.
Two controls are weighted double because the logs justify it. A real focal length range beyond 85mm, and an aperture with a placeable focus plane. Those two together can lift a score by four points, which is how Moss And Marble reaches eight with no time-of-day control at all.
Nominal controls score zero. A time-of-day slider that produces forty minutes of slightly warmer grey is not a time-of-day control, and Causeway is marked accordingly.
The harder and more arguable of the two scores. It assesses what the world supplies before any camera is involved: how many distinct lighting conditions occur, whether weather arrives, whether terrain separates into readable planes, whether anything moves in a way that rewards a slow shutter.
It is not an assessment of art direction, and a game can be beautiful and score badly. Causeway's continent is handsome and scores four, because handsome is not the same as photographable and one flat sky over everything is a photographic problem regardless of how well it is painted.
This score is awarded last, after the kept ratio is known, which risks circularity. I have not found a way to avoid that without pretending the log does not exist, and the risk is recorded here rather than hidden.
Every game is purchased at retail with the publication's own money. No review copies, no early access, no press builds, and nothing reviewed before release. Games are played on one machine at one fixed set of graphics settings, chosen once and never adjusted for a screenshot.
That last point matters. A frame that only works at maximum settings is a specification rather than a composition, and this publication has no interest in comparison shots, hardware testing, or images produced by configurations no player would use.
One reviewer, which is the largest weakness. Every kept frame reflects one person's taste, and a viewer who prefers dense maximal images would produce a different table in the essay from the same logs. The test — a second person reviewing the logs blind — has not been run.
Small samples. Fourteen games across three issues, and several findings rest on three or four cases. Nothing here should be read as a claim about the medium.
No modded games, no reshade, no injected tools of any kind. Every frame is what the game produces unaided, which excludes a large and skilled community and is a genuine limitation rather than a principle.
No multiplayer. Photography in a game full of other people is a different activity with different constraints, and none of the procedures above would survive contact with it.
Wrong figures are struck through with the replacement and the date beside them. Wrong findings stay on the page with the reasoning that overturned them written underneath: In Defence Of The Almost Empty Frame has been narrowed once and says where and why.