TOPIC:

In "Text to match" I have complete alphabet: ABCDEFGHIJKLMNOPQRTSTUVWXYZabcdefghijklmnopqrtstuvwxyz0123456789()+=!?,.

However, in Online search in Google fonts, it says "Matching 'ABCDEFG"', which means it's only looking at the capitals. And indeed if I place some lowercase chars to the first 7 characters of my match text, it finds different matches!

I realize it may be slower, but I need it to work on all the characters. Precision is more important to me than speed. I could not find anywhere in preferences how to increase this limit.

Hi Dmitry,
yes you are right, the online matching is limited to 7 characters and this cannot be changed in application preferences.
There is a number of reasons for this limit:
a) It's faster for you
b) Each additional character will take more time on our matching server, which means it will slow down other users matching requests
The obvious solution to (b) is to add more matching servers on our side, but this is a rather expensive solution and (more important) the servers will remain idle most of the time, because the average Find my Font user uses 3-4 letters and this is usually enough for an accurate font identification.

The trick for an accurate matching is to select the most distinct (=the most "strange") letters of the available image text.
Follow these hints:
1. If you have an image containing the whole alphabet, try these letters first: QRSgrkt
They usually return the best matching results

2. If you have a script text which begins with a capital letter with many flourishes, try to select the 2nd, 3rd letters etc. (avoid the first) and try also to include the last one.
The reasoning behind this is:
(a) An initial capital letter with many flourishes is likely to be an open type alternate glyph and Find my Font don't match all alternate glyphs.
(b) The intermediate letters of a connected (script) text usually overlap, while the last letter's tail is visible which makes it a better matching candidate.

In my experience the 7 letters are enough to accurately match any font.
If you get different matching results by selecting different letters, chances are:
i) We don't have this font in our online database
ii) Your image contains very ordinary (not distinct) letters or the font itself is very ordinary and similar to many other fonts
iii) The resolution of your image is really low (=letter height is less than 15 pixels). In this case try the "Image Correction and Deformation Tool" using the [x] Enlarge option, before selecting any letters.

I hope you will find the above hints useful and they will improve your matching results

Thanks for the tips. I remain unconvinced, however. 7 characters may be OK when you're simply identifying a font, i.e. looking for an exact match in the entire database. My use case is different: I'm searching for lookalike fonts among the free Google fonts, not exact matches. I'm pretty sure that for this, more characters would improve the quality of the matches.

You have good technology, why do you artificially limit its usefulness? Why not have the 7 chars limit for the free trial but remove it in the Pro version?

As for your servers being idle, have you looked into AWS lambdas? There, you only pay for the actual usage, removing any scalability issues.

Thanks, we'll consider your thoughts, although our code requires a dedicated server - it's not appropriate for AWS lambdas.
You are right, if you are searching for lookalikes, more characters will improve the matches.

If your main interest is the free Google fonts, the only way to achieve this at the moment is:
a) Download all Google fonts in your hard disk (I think they offer either a zip download or you can create a repository and update the fonts periodically).
b) Go to Find my Font menu "Fonts" => "Fontsets" click on "Add Fontset" and select your local Google Fonts folder
c) Select the new Fontset and click on "Create/Update Index", to make the matching about 100x times faster

If you use the above method you can match any number of letters you want (even the whole alphabet) because there is no letter number limit in local matching.