Subject:
|
Re: Chrome Parts
|
Newsgroups:
|
lugnet.cad
|
Date:
|
Fri, 9 Aug 2002 00:04:35 GMT
|
Viewed:
|
728 times
|
| |
|
|
In lugnet.cad, Travis Cobbs writes:
> In lugnet.cad, Steve Bliss writes:
> > I generate them with LDLite, scaled up 3x, with thick edges. Then I run
> > them through ImageMagick to reduce by 3x (and convert to gif). Finally, I
> > 'batch convert' them with Paint Shop Pro, because ImageMagick doesn't
> > compress the gifs (I assume there's a patent/licensing issue between Unisys
> > and ImageMagick).
> >
> > It's the 3x reduction that blends the dithered pixels, resulting in smooth
> > colors. The dithering was supposed to give some indication of the chrome-ness.
>
> Looking at the images, it doesn't appear to matter much, but you might get
> better results if your initial rendering were at 4x (or possibly even 2x).
> The reason for this is that a 3x3 sample of a checkerboard pattern will
> result in two slightly different shades. For instance, if the two blended
> colors are blue and green, one pixel will sample 4 blue and 5 green, while
> the pixel next to it will sample 5 blue and 4 green. If you did 2x or 4x
> rendering, all the pixels would line up with the checkerboard and they would
> all sample to exactly the correct color.
Which leads us back to ldglite. Since I let opengl do the linewidths
it supports 2x or 4x or even 2.5x (any floating point will do).
On the other hand, while I think you'll be satisfied with the gold and
silver colors, I'm not sure I implemented the other dithered colors
right. Which colors are supposed to be dithered, and is there really
any reason to dither them?
Don
|
|
Message has 1 Reply:
 | | Re: Chrome Parts
|
| (...) The linewidth isn't important, I just use it in conjunction with the scale up/down to get anti-aliasing and lines with normal widths. (...) None of the standard/extended colors are *supposed* to be dithered. Only the non-standard colors (...) (22 years ago, 9-Aug-02, to lugnet.cad)
|
Message is in Reply To:
| | Re: Chrome Parts
|
| (...) Looking at the images, it doesn't appear to matter much, but you might get better results if your initial rendering were at 4x (or possibly even 2x). The reason for this is that a 3x3 sample of a checkerboard pattern will result in two (...) (22 years ago, 8-Aug-02, to lugnet.cad)
|
25 Messages in This Thread:
 
 
- Entire Thread on One Page:
- Nested:
All | Brief | Compact | Dots
Linear:
All | Brief | Compact
This Message and its Replies on One Page:
- Nested:
All | Brief | Compact | Dots
Linear:
All | Brief | Compact
|
|
|
|
