To identify a typeface with FontExpert, you have to cut out single characters of a scanned document to be analysed by the software. FontExpert can recognize the characters A-Z in upper and lower case as well as the numerals 0-9.

Step 1:

First, the printed document has to be scanned using the scanning software at hand, which in the most cases will be a Photoshop PlugIn. Within the graphics software (e.g. Adobe Photoshop^®, Paint Shop Pro^TM, etc.) the scan will be prepared.
Sloped text has to be levelled horizontally. The color depth should be decreased, because FontExpert internally converts the characters to black & white anyway. Also, the transfer time on the web will increase with high color images.
It is recommended to decrease the color depth to black & white.

A few remarks on the scanning: The better the quality of the printed document, the more exact the identification results will be. The more characteristic the examined letters are, the more definite the result will be. You should choose a resolution for scanning that leads to character sizes of 200 to 600 pixels. The smaller the document, the higher the resolution which should be chosen.
The following values may be taken as a rule of thumb: If the character height is less then 0.3 inch, a resolution of 1200 dpi should be chosen, for bodytext of 8 point even 2400 dpi. For character sizes up to 0.6 inch, a resolution of 600 dpi is recommended, for larger characters 300 dpi are sufficient.

Step 2:

Then, select a single, isolated character using the magic wand or e.g. the lasso tool (you should always choose the most characteristic letters from the scanned text - usually, characters like g, G, R are a better choice than i or l).
Select, if possible, characters which are not toosmall. This is much easier to identify then this .
Copy the selected charatcer to the clipboard (Edit/Copy). Then, open a new image by selecting File/New. In Photoshop the image size is already the size of the character in the clipboard, so you can just click OK. Then, insert the selected character from the clipboard (Edit/Paste). There may be white space around the character, it will be filtered by FontExpert.

Find some samples of good and bad images here.

Step 3:

Save this new image onto your harddisk (File/Save as... or File/Save copy as...). Choose the file type GIF (FontExpert can also read JPEG- or Windows-BMP-files, but GIF images lead to the best results). Name the file, e.g. "B.gif".

Repeat this for 1 or 2 further characters. Usually 2 or 3 characters will lead to reliable results.

Step 4:

Now, go to the website "Online-Font-Identification". You will find an input form for up to 5 character image files. Enter the characters and the image file names which you prepared for the identification procedure.

Step 5:

Click the button "Start font identification".

After the research FontExpert will show a result page, or it gives an information that no matching fonts could be found.

Search Result

The five best matching typefaces will be shown as bitmap samples.
If the bitmap images of the fonts look very much alike the the typeface on your document, you can access the details of the result including font names and type foundries by clicking the button "Detailled Result".

For showing the detailed report we charge US$ 5.

After taking your credit cards details by using a secure connection (SSL), the details will be shown.

If FontExpert find more matching fonts by other foundries, these alternatives will also be listed. By clicking the foundry name link you will get more information about the type foundry and their vendors.

If FontExpert does not find any matching fonts, there may be different reasons:

• maybe the characters in the image files are not properly prepared for the FontExpert identification algorithm
• the characters may be too small
• if you have selected only one character, please try one or two more
• if you can select more characteristic letters, try these
• the entered character is not equal to the one in the image file
• overlapping characters are sometimes difficult isolate correctly
• colored images may be converted incorrectly (e.g. inverted)
• if the color depth of the character image is e.g. 16 millionen colors, the image file can get too large. Please reduce thze color depth.
• if the has not been scanned horizontally, the characters may be sloped und therefore not justified correctly
• the font may have been modified manually. In this case, FontExpert has no chance to identify the original typeface.
• it is possible that the typeface is not yet in the FontExpert database. If you find out the name of the font by other means than FontExpert, we would be thankful if you let us know, as we wish to keep our font database as complete and up-to-date as possible.

Samples of good and bad images for the font identification:

	bad	The automatic reduction of the color depth from 16 to 2 colors does not reproduce the correct character.
	good	This image is already black & white. The font can be identified exactly.
	bad	FontExpert needs single characters for identification. Complete words or phrases cannot be recognised.
	good	This is a black & white image with the right size to be easily identified.
	bad	FontExpert can only work on black characters on white background. Is the colors are inverted before, the font will be identified.
	bad	This image should also be changed to black & white.
	bad	This character is too blurred to be identified correctly.
	good	Good image in a good size, can be identified without problems.
	bad	This image has been scanned with very low resolution. The character height is only about 40 pixels. This is not enough for a good identification.
	bad	This character is also much too small.
	bad	This character is not isolated. The black bar on the right make is impossible to identify ther font.
	bad	This character is not complete and cannot be identified.