Filter Finder

Filter Finder

Compare filter lines from every brand on one set of rated axes, and find the closest substitute for the one you cannot get.

Table of Contents

• What it is for
• Opening it
• One tile is one filter line
• The nine rated axes
• Narrowing the catalogue
• Tiles, list and sorting
• Reading one line
• Closest matches
• Comparing up to four lines
• Exporting a PDF
• Saved views and shareable links
• Where the ratings come from
• Using it with a project
• What it does not tell you
• Related Documentation

What it is for

Every manufacturer describes its own diffusion in its own words. Deciding between a Black Pro-Mist and a Glimmerglass normally means reading two marketing pages written to different standards and guessing at the difference.

The Filter Finder rates every line in the catalogue on one shared set of axes, so the comparison is like for like. It answers two questions that come up constantly in prep:

• What does this filter actually do to the image?
• The rental house does not have it — what is the nearest thing they do have?

Everything runs offline in your browser. Nothing is sent anywhere.

Opening it

Additional Tools → Filter Finder in the sidebar. The ? button on the page starts a six-step guided tour.

One tile is one filter line

A tile is a line — Black Pro-Mist, CineBloom, ARRI FSND — not a single strength. Pick the line first; the strengths it is sold in (1/8, 1/4, 0.9, "2–8 stops") are listed on the line itself, taken from the maker's published range.

This matters when searching. Typing 1/4 finds every line sold in a quarter strength rather than one product called "1/4".

The nine rated axes

Every line is rated on the same nine axes:

Each is graded on one five-step scale — None, Minimal, Low, Moderate, Strong — or marked Not applicable where the axis has no meaning for that class. A polariser has no halation rating because halation is not a thing a polariser does; that is different from rating it None, and the tool keeps the two apart.

Three of the axes — resolution softening, colour shift and density accuracy — are the ones that matter for light-control glass (ND, IRND, polarisers, variable ND), where the question is not what the filter adds but what it costs you.

Colour direction is not one of the nine. Which way a cast goes — none, warm, and so on — is a fact about the line rather than a strength, so it has no ladder and no rating. It appears in the comparison table alongside brand, effect class and available sizes, not in the ratings.

Narrowing the catalogue

The search box takes a name, a brand, an effect class or a strength. Alongside it:

• Brands — 13 of them, including a Generic entry for unbranded rows such as "ND Grad SE", which exist so a gear list that just says "Rota-Pola" still resolves to something.
• Effect classes — 19, from Contrast Control and Black Halation through Neutral Density, IRND, Variable ND, Polariser and Protection & Clear.
• Available in this size — every size the catalogue records, so you can ask what exists in 4×5.65 before you ask what it does.
• Sold in this strength — every published strength, grouped so the four different vocabularies (fractions, densities, millimetres, stop ranges) do not interleave.
• Manufacturer-rated lines only — hides the in-house-rated generic rows.
• Only lines I own — matched against your gear list.

The axis facets work in both directions

Each of the nine axes has one control offering both ≥ and ≤:

• At least this much — ≥ Moderate on skin smoothing finds the lines that do a lot of it.
• At most this much — ≤ None on resolution softening finds the lines that are genuinely clean on that axis.

The second direction is the one people actually need for four of the nine. Nobody looks for a filter with at least this much colour shift — that is the thing you are trying to avoid. ≤ None is a real setting and is the only way to ask which lines cost you nothing on an axis.

Below tablet width the whole band folds into a Filters sheet, with the number of active facets shown on the button. Clear filters resets everything; the header always shows how many of the total lines are currently shown.

Tiles, list and sorting

Tiles is the browsing view. List is the working view: one row per line, with columns for brand, effect class, grades and all nine axes. Columns can be rearranged and hidden — the filter-line column always stays, because a row with every identifying column switched off is an unreadable grid of adjectives.

The sort menu offers the line name both ways, brand and effect class A–Z, and each axis strongest-first. The axes sort one way only on purpose: "show me the weakest halation" is not a question anyone asks. The list view's own column headers still sort both ways.

Reading one line

Selecting a line opens its detail panel:

• What it does — the nine ratings, on the shared scale.
• Strengths — the grades the maker publishes for that line.
• Available sizes — every recorded size, each one clickable through to that exact size.
• Closest matches — see below.
• What these ratings were read from — the source, named.

A line that has been announced but is not shipping yet is labelled as such rather than quietly listed as though you could order it.

Closest matches

This answers the substitute question. Lines are ranked by how closely their ratings agree, and the result says so in numbers rather than adjectives: "7 of 9 axes compared, biggest gap: halation", or "identical on all of them".

Two things are deliberate here. Only axes that both lines actually rate are counted, so a comparison is never inflated by axes one of them treats as not applicable. And the ranking stays inside sensible company — asking for a substitute for a variable ND returns other light control, never a mist, because a filter you could not physically put in the same tray is not an answer.

If nothing in the catalogue shares an axis with the line, the tool says that instead of offering the least-bad match.

Comparing up to four lines

Add lines to the shortlist and open the comparison to see two to four side by side, axis by axis, with the differences called out. One line is not a comparison — that is the detail panel, which is better at it and one click away — so the table is not offered until you have picked two.

Below the ratings the table carries the facts that have no ladder: brand, effect class, grades, available sizes, colour direction, and Blacks — which restates shadow lift in the polarity a filter chart uses, and is marked as derived so you are not told the same thing twice as if it were two independent findings.

Exporting a PDF

Two exports, both from where you are already looking:

• The comparison — the shortlist table as a PDF, from the shortlist tray or the comparison header.
• A filter sheet — the selected line on its own: ratings, published strengths, sizes and the wording they were read from, on one page you can hand to a rental desk without first inventing a second line to compare it against.

Both use the same projection and the same drawing code, so the two can never disagree.

Saved views and shareable links

Copy link puts the current state — search text, facets, selection, shortlist — into the URL, so a link reproduces exactly what you are looking at for whoever you send it to.

Saved views keeps a named set locally ("Berlin job — mist tests") for the ones you come back to. Saved views live in your browser; a link is what you share.

Where the ratings come from

Two sources, and the tool always says which one a line uses:

• Manufacturer — read off the maker's own published wording for that line.
• In-house reference matrix — the cross-brand comparison used for generic type rows, which stand in for the unbranded entries a gear list often contains.

Where a maker publishes a limitation, it travels with the line rather than being smoothed away. Clear glass, optical flats and UV are in the catalogue rated None on all eight axes that apply to them rather than left unrated — density accuracy is the ninth, and it is Not applicable, because a filter with no density has none to be accurate about. That is the claim those lines make, and it turns them into the baseline every diffusion line is implicitly read against.

Using it with a project

In Project Requirements → Mattebox & Filter, a filter row can optionally be linked to a line from this catalogue. The row works without it; the link adds the nine ratings, the published strengths and the available sizes directly under the row, so the person reading the requirement does not have to go and look the filter up.

What it does not tell you

• No variable ND is rated above Low for density accuracy, because no maker in that class publishes one — the manufacturers' own word for the range is "approximate".
• Two protection filters compared side by side report not comparable rather than a confident 100%. They differ in what light they block, and these nine axes do not measure that.
• Ratings describe what a line does, not whether it suits a face, a lens or a look. They narrow the shortlist you test; they do not replace the test.

Related Documentation

• User Guide — Tools & Simulators
• Camera Tools
• Device Library

---

Last Updated: 2026-09-10 Version: 0.790.2