CompuServe Thread

#ANI-PRO Font Editor

9 messages in this thread
#36997From: Paul LindMar 10, 1993 7:05 PM
Phil, There really isn't such an item as Anti-aliased fonts. In an extremely simple explanation, programs use standard font files and use a "dirtying" routine to create blurred edges by calculating pixels with colors in between the font color and the background color. This blurring allows the text to look "cleaner" when in actuality it is performing the exact opposite function. This is the manner in which programs along the lines of Grasp, TypeRight, etc. achieve anti-aliasing. There is no trick font code. Before I get attacked by others let me state that this is an EXTREMELY basic explanation, and I'm purposely avoiding discussing sampling mechanics and/or the usage of subpixel increments in performing anti-aliasing. In Ani Pro 1.3 you can achieve a similar result by using the Unzag ink and applying it to a screen of text. It's quick and dirty but it performs the same function. (Check out page 2-47 of the Ani Pro manual regarding Unzag.) If you have used the Show'em poco routine and played with the anti-aliasing option then you are actually using the unzag routine, I've outlined. There may be other suggestions from members on the forum regarding anti-aliasing. It's essentially the same whether you're working with text or images. Hope this helps, Paul
#37012From: doug wolfgramMar 10, 1993 9:05 PM
Paul, Wrong!!!! There is no auto-anti-aliasing in GRASP. What Phil is referring to is a technique he developed at AdWare which uses as many as 4 specifically developed fonts which are hand-anti-aliased. Unzag does a horrible job. Of course, you are right in that true anti-aliasing requires 24 bit graphis to do properly. What Phil is suggesting is a routine similar to GRASP's where we can create several fonts and have a POCO routine overlay them like we do in GRASP. Our biggest problem now is anti-aliased fonts in Windows.
#37034From: Paul LindMar 11, 1993 12:33 AM
Doug, If Phil works with the technique in Grasp, then why doesn't he write a poco routine to utilize them in Ani Pro? I have a copy of Grasp but never really use it because it doesn't do anything that I need. I apologize for not digging out the program and researching before I replied. I was trying to present the fact that if Grasp did anti-alias text then it would have to use some method of sampling or a step procedure. TypeRight does use that technology. It seems like a virual impossibilty to "anti-alias" text without using some type of stepping procedure on a screen subpixel level. If Phil is combining Fonts then the text isn't antialiased at all. It's overlayed in some manner . I'd be interested in how a font combination takes account of background color. What if you want to antialias text over a image? How does this multiple font technology take into account the multiple colors of the background image? Anti-aliasing text is not hard to do but I've never seen it approached from that direction. If you wanted to antialias text why not simply write the code to actually do it? Does Phil's overlay technique allow for scaling of the fonts? I'd love to see an example of the implementation. I've spent years writing proprietary font code, and we specifically design fonts for a large number of Graphic software companies. You can antialias text in 8 bit, it just slows down the average program and limits the number of colors you can use elsewhere. Whether or not the Unzag Ink is great, it is performing an actual type of independent antialiasing by smooting the extremes of the stair case effects. It will work on any object by taking into account the object color and background color. Once again combining multiple fonts is not anti aliasing. Who actually creates these multiple fonts? How long does it take? Do they create full character sets? It usually takes me close to a month to properly lay out the design typography, calculate weight,kerning & leading, and digitally create a single font, and we run one of the fastest type foundries that I know of. If I had to create the font four times to get a single pseudo anti aliased font, I think I'd devote the time to write the antialiasing code. Paul
#37041From: Phil ShatzMar 11, 1993 4:39 AM
Ok so I used the wrong term. I'm talking about overlaying rasters on top of each other so as to create a smooth font. This technique takes no account of the background color. For our presentation work 95% of the time the background is dark enough that this doesn't matter. This technique doesn't allow for any scaling whatsoever. Its a totally manual process. We use the ancient font editor that Doug wrote to edit our GRASP fonts. Some people use Fontrix others you a new tool called PCXFNT thats available over in GO GRASP. We can create a set of small antialiased fonts in 3-5 days using these tools. Since they are only for onscreen display and don't need to be scaled we don't have all the constraints that you do in laying out vector based fonts. I'll email you a sample.
#37051From: doug wolfgramMar 11, 1993 9:37 AM
Thanks for the clarification Phil.
#37066From: Paul LindMar 11, 1993 10:40 AM
Phil, I checked out your example and it is really a strange way of presenting a font. I'm taking it for granted that as I hit the enter key, it is overlaying the four layers of matrix. In the first layer the font is purposely missing pixels to make it appear to look worse than it normally would. ( in certain characters angled pixels are missing that wouldn't be in a standard font). By the third layer you are essentially achieving the same level of display as you would with a noraml screen font. And the fourth level is simply shadowing the right side of each letter to make the letter appear to be offset from the screen. This tends to seperate the text from the underlying background or image but quickly breaks down when applied over a background. We have used similar techniques in creating screen fonts for various programs. It no doubt has a purpose, but should in no way be confused with anti aliasing. If that font was antialiased you wouldn't have to fake a right shadow to make it look nice. Not to mention the fact that an antialiased 8 bit font of that resolution would contain no less than 10 colors to properly "smooth" the text. It should be a breeze to recreate the effect in Ani Pro. I'm sending you a gif file of an actual postscript antialiased font. Look at it and you'll quickly be able to tell that the font is extremely smooth and doesn't need a shadow. Th text is red on white to force a strong antialias. Notice how you don't see any of 17 various shades of red on the border. The file was done entirely in Ani Pro. Since we've determined that the other type of fonts are not scalable, look rough over multicolored background images, and use shadows to fake screen seperation. I have only a few more questions. How do the fonts work when you are jumping screen resolutions. (Do they work equally well with both 640X400, 640X480, 800X600, 1024X768?) If I want to change the color of the text to either black or white, is there an algorithm to automatically readjust all of the rest of the layers? [>> Continued in next msg]
#37067From: Paul LindMar 11, 1993 10:40 AM
[>> Continued from previous msg] Sorry about all of the questions, you have to understand that a large part of my life is centered around typography. I've been designing type on varying levels for over 10 years, and I sit in a room with hundreds of volumes of font code. Sad but true !!!! Hey, someone has to do it!!!! Paul
#37052From: doug wolfgramMar 11, 1993 9:37 AM
Paul, You are right. But when working with a limited medius as we have done since CGA days, we have to come up with clever tricks to improve the quality of our presentations. GRAFX does not create presentations without anti-aliased text any more (except in Windows!). 'Anti-aliased' is misused here as you state because true anti-aliasing is a sub-pixel blend between adjacent colors and with complicated colors, this is very complex unless you are in 24 bit mode. In any case, I am sure this discussion has educated some of our new listeners. Thanks!
#37040From: Phil ShatzMar 11, 1993 4:39 AM
Thanks! My point exactly..