ftp.xmission.com/2014.06.ftp.xmission.com.tar/ftp.xmission.com/pub/lists/fractdev/archive/fractdev.200108

Internet Message Format | 2001-08-29 | 19KB

From: "Jonathan Osuch" <[e-mail]>
Subject: (fractdev) Rotate and Skew
Date: 01 Aug 2001 20:40:39 -0500

Tim,

That's not meant as an insult.  8-))

It appears from a recent FOTD from Jim Muth that the cvtcentermag() routine
(in miscres.c) is not recalculating the same rotation and skew values when
an image is zoomed (without changing the rotation or skew).  There is a
small difference that is noticeable even from the default mandelbrot.  This
problem was also noticed by Lee Skinner some time ago with one of his deep
zoom images.  So this problem occurs with the ap-math as well.

Problems occur when an image is saved because the corner values are saved,
and not the center mag values.  Since the PARs now default to saving the
center mag values, there can be a difference in both rotation and skew
between the PAR and the GIF.

Thoughts?

Jonathan



Thanks for using Fractdev, The Fractint Developer's Discussion List
Post Message:   [e-mail]
Get Commands:   [e-mail] "help"
Administrator:  [e-mail]
Unsubscribe:    [e-mail] "unsubscribe fractdev"


-------------------------------------------------------------------------------

From: "Jonathan Osuch" <[e-mail]>
Subject: Re: (fractdev) Rotate and Skew
Date: 02 Aug 2001 20:30:28 -0500

Tim,

The problem disappears when corners= is used to force the use of the corners
instead of center-mag.

It is possible the problem arises because the rotation and skew values are
being truncated to three significant digits before being put in the PAR.
I'll check.

Jonathan



Thanks for using Fractdev, The Fractint Developer's Discussion List
Post Message:   [e-mail]
Get Commands:   [e-mail] "help"
Administrator:  [e-mail]
Unsubscribe:    [e-mail] "unsubscribe fractdev"


-------------------------------------------------------------------------------

From: Tim Wegner <[e-mail]>
Subject: Re: (fractdev) Rotate and Skew
Date: 02 Aug 2001 21:07:07 -0600

> It is possible the problem arises because the rotation and skew values are
> being truncated to three significant digits before being put in the PAR.
> I'll check.

That sounds plausible. The logic that determines how many digits
are written out is in miscovl.c if memory serves.

Rotating and skwing put a lot of stress on the precision because of
dividing by small numbers. It is probably possible to rearrange the
algebra in an alternative way that is more stable.

I can look at it this weekend if it is still needed. My son is visiting
now but is leaving saturday morning.

Tim


Thanks for using Fractdev, The Fractint Developer's Discussion List
Post Message:   [e-mail]
Get Commands:   [e-mail] "help"
Administrator:  [e-mail]
Unsubscribe:    [e-mail] "unsubscribe fractdev"


-------------------------------------------------------------------------------

From: "Jonathan Osuch" <[e-mail]>
Subject: Re: (fractdev) Rotate and Skew
Date: 03 Aug 2001 18:02:15 -0500

>> It is possible the problem arises because the rotation and skew values
are
>> being truncated to three significant digits before being put in the PAR.
>> I'll check.

> That sounds plausible. The logic that determines how many digits
> are written out is in miscovl.c if memory serves.

I increased the significant digits to six and the problem went away.  I'll
need to check Lee Skinner's example to see if that is sufficient.

Jonathan



Thanks for using Fractdev, The Fractint Developer's Discussion List
Post Message:   [e-mail]
Get Commands:   [e-mail] "help"
Administrator:  [e-mail]
Unsubscribe:    [e-mail] "unsubscribe fractdev"


-------------------------------------------------------------------------------

From: Tim Wegner <[e-mail]>
Subject: Re: (fractdev) Rotate and Skew
Date: 04 Aug 2001 10:09:49 -0600


> I increased the significant digits to six and the problem went away.  I'll
> need to check Lee Skinner's example to see if that is sufficient.

We could simply increase digits across the board, or if that is too
ugly (and unnecessary because the number of digits is carefully
tuned to work right for most circumstances - e.g. normal zoom
boxes) we could detect situtaions when more are needed. A really
simple way would be to have any sort of skewed zoom box causing
more digits too be used.

OTOH if the additional digits in PAR files aren't too annoying, just
do it in all cases.

Tim


Thanks for using Fractdev, The Fractint Developer's Discussion List
Post Message:   [e-mail]
Get Commands:   [e-mail] "help"
Administrator:  [e-mail]
Unsubscribe:    [e-mail] "unsubscribe fractdev"


-------------------------------------------------------------------------------

From: Rich <[e-mail]>
Subject: Re: (fractdev) Rotate and Skew
Date: 06 Aug 2001 00:31:35 -0600


In article <3B6BC9ED.12168.2CB450@localhost>,
    Tim Wegner <[e-mail]>  writes:

>
> OTOH if the additional digits in PAR files aren't too annoying, just
> do it in all cases.

I like this.
--
Ask me about my upcoming book on Direct3D from Addison-Wesley!
Direct3D Book <http://www.xmission.com/~legalize/book/>
         Home <http://www.xmission.com/~legalize/>
    Fractals! <http://www.xmission.com/~legalize/fractals/>

Thanks for using Fractdev, The Fractint Developer's Discussion List
Post Message:   [e-mail]
Get Commands:   [e-mail] "help"
Administrator:  [e-mail]
Unsubscribe:    [e-mail] "unsubscribe fractdev"


-------------------------------------------------------------------------------

From: "Jonathan Osuch" <[e-mail]>
Subject: Re: (fractdev) Rotate and Skew
Date: 06 Aug 2001 20:26:36 -0500

Rich,

>> OTOH if the additional digits in PAR files aren't too annoying, just
>> do it in all cases.

> I like this.

That's what I intend to do.  I've changed the values from x.xxx to x.xxxxxx,
but is that enough to cover all cases.  I'm not even sure I know how to test
it.

Jonathan



Thanks for using Fractdev, The Fractint Developer's Discussion List
Post Message:   [e-mail]
Get Commands:   [e-mail] "help"
Administrator:  [e-mail]
Unsubscribe:    [e-mail] "unsubscribe fractdev"


-------------------------------------------------------------------------------

From: "Jonathan Osuch" <[e-mail]>
Subject: (fractdev) WinCVS
Date: 06 Aug 2001 20:29:09 -0500

Tim,

I played with WinCVS over the weekend.  It almost works using a local
repository.  I must still be missing something.

Jonathan



Thanks for using Fractdev, The Fractint Developer's Discussion List
Post Message:   [e-mail]
Get Commands:   [e-mail] "help"
Administrator:  [e-mail]
Unsubscribe:    [e-mail] "unsubscribe fractdev"


-------------------------------------------------------------------------------

From: Tim Wegner <[e-mail]>
Subject: Re: (fractdev) WinCVS
Date: 06 Aug 2001 21:56:06 -0600

Jonathan wrote:

> I played with WinCVS over the weekend.  It almost works using a local
> repository.  I must still be missing something.

The command line CVS is not that hard to use. I suggest we start
with that. It's available when you ssh into fractint.org. If we ever get
WinCVS working, we can instantly use our work with the command
line version.

One advantage of using the command line version is you get to
understand the concepts. Then WinCVS makes more sense.

Tim


Thanks for using Fractdev, The Fractint Developer's Discussion List
Post Message:   [e-mail]
Get Commands:   [e-mail] "help"
Administrator:  [e-mail]
Unsubscribe:    [e-mail] "unsubscribe fractdev"


-------------------------------------------------------------------------------

From: Rich <[e-mail]>
Subject: Re: (fractdev) WinCVS
Date: 06 Aug 2001 22:58:30 -0600


In article <001501c11ee0$5d608a00$2c0e270c@laptop20>,
    "Jonathan Osuch" <[e-mail]>  writes:

> I played with WinCVS over the weekend.  It almost works using a local
> repository.  I must still be missing something.

What are the things that don't work?  I've never had a problem with
either a local repository or with a unix server repository.  (Apparently
the NT server one is a little flakey last I heard.)
--
Ask me about my upcoming book on Direct3D from Addison-Wesley!
Direct3D Book <http://www.xmission.com/~legalize/book/>
         Home <http://www.xmission.com/~legalize/>
    Fractals! <http://www.xmission.com/~legalize/fractals/>

Thanks for using Fractdev, The Fractint Developer's Discussion List
Post Message:   [e-mail]
Get Commands:   [e-mail] "help"
Administrator:  [e-mail]
Unsubscribe:    [e-mail] "unsubscribe fractdev"


-------------------------------------------------------------------------------

From: "Jonathan Osuch" <[e-mail]>
Subject: Re: (fractdev) Rotate and Skew
Date: 08 Aug 2001 20:38:13 -0500

Tim,

Take the following PAR, start it, zoom all the way out once.  Let the image
generate, save it, and then make a PAR from it.  Run the PAR, set debug=50,
and then reload the saved image.  You should see a pixel pattern that would
indicate that one of the images is shifted.  Rotated at least.  This doesn't
happen if you zoom in.  At least not until you get to the ap-math.

Jonathan

bob_and_weave      { ; ;
                     ;  Jonathan Osuch [e-mail]
                     ;  Aug 08, 2001 at 20:22:50
                     ; Version 2001 Patchlevel 13
  reset=2001 type=mandel
  center-mag=-0.99428940396649250/+0.29970533391439380/1.276515e+011/1/-37\
  .4999/-14.9313 params=0/0 float=y maxiter=1000 inside=0 sound=off
  }




Thanks for using Fractdev, The Fractint Developer's Discussion List
Post Message:   [e-mail]
Get Commands:   [e-mail] "help"
Administrator:  [e-mail]
Unsubscribe:    [e-mail] "unsubscribe fractdev"


-------------------------------------------------------------------------------

From: Tim Wegner <[e-mail]>
Subject: (fractdev) C700+MASM
Date: 15 Aug 2001 23:12:57 -0600

IJonathan,

I'm trying to get C700 and MASM installed on a laptop. I think I
have C700 OK, at least I can do hello world. Do you remember any
special incantations in autoexec.bat or config.sys needed for
MASM under Windows Millenium Edition?

I've gotten it running under Win98 on my regular machine.

I'd much rather have Win98 on the laptop, but WinME is what it
came with.

Geesh this development environment seems 1000 years old, and it
only dates to about 1992!

Tim

Thanks for using Fractdev, The Fractint Developer's Discussion List
Post Message:   [e-mail]
Get Commands:   [e-mail] "help"
Administrator:  [e-mail]
Unsubscribe:    [e-mail] "unsubscribe fractdev"


-------------------------------------------------------------------------------

From: "Jonathan Osuch" <[e-mail]>
Subject: Re: (fractdev) C700+MASM
Date: 16 Aug 2001 19:31:12 -0500

Tim,

> I'm trying to get C700 and MASM installed on a laptop. I think I
> have C700 OK, at least I can do hello world. Do you remember any
> special incantations in autoexec.bat or config.sys needed for
> MASM under Windows Millenium Edition?

I didn't have to do anything special.  My autoexec.bat contains the
following lines:

SET
PATH=C:\WINDOWS;C:\WINDOWS\COMMAND;C:\UTILITY;C:\SPRINT;C:\DJGPP\BIN;C:\C700
\BIN;C:\MASM\BIN;C:\MASM\BINB;C:\TCL\BIN
SET LIB=C:\C700\LIB;C:\C700\MFC\LIB
SET INCLUDE=C:\C700\INCLUDE;C:\C700\MFC\INCLUDE;C:\MASM\INCLUDE
SET HELPFILES=C:\C700\HELP\*.HLP;C:\MASM\HELP\*.HLP
SET INIT=C:\C700\INIT;C:\MASM\INIT

I copied them from the autoexec.bat on the old laptop before I had to
destroy all the data on it to get Win98 to work again after removing the
Linux partition.  I would guess that only the path and include entries are
necessary.

Yes, the environment is ancient.  But, changing involves a major effort.

Jonathan





Thanks for using Fractdev, The Fractint Developer's Discussion List
Post Message:   [e-mail]
Get Commands:   [e-mail] "help"
Administrator:  [e-mail]
Unsubscribe:    [e-mail] "unsubscribe fractdev"


-------------------------------------------------------------------------------

From: Tim Wegner <[e-mail]>
Subject: Re: (fractdev) C700+MASM
Date: 17 Aug 2001 16:06:58 -0600

Jonathan wrote:

> I didn't have to do anything special.  My autoexec.bat contains the
> following lines:

I got it working. I had to rediscover a copy of points:

1. We're using a different version of nmake
2. For some reason loading dosxnt.386 was left out of
\windows\system.ini

For future reference, copying \c700\bin is actually a good idea
(picks up the right nmake). Then if autoexec.bat, config.sys, and
windows.ini have the right magic, all is well.

I'm off to dive into rush hour to catch my plane. Will wave when I go
over Iowa :-)

Tim


Thanks for using Fractdev, The Fractint Developer's Discussion List
Post Message:   [e-mail]
Get Commands:   [e-mail] "help"
Administrator:  [e-mail]
Unsubscribe:    [e-mail] "unsubscribe fractdev"


-------------------------------------------------------------------------------

From: "Jonathan Osuch" <[e-mail]>
Subject: (fractdev) Autologmap bug
Date: 18 Aug 2001 21:12:53 -0500

Tim,

Did you ever notice that the autologmap routine misses color #2?  This
problem goes back to the changes I did for calculating the logmap on the fly
(version 19.2).

Jonathan



Thanks for using Fractdev, The Fractint Developer's Discussion List
Post Message:   [e-mail]
Get Commands:   [e-mail] "help"
Administrator:  [e-mail]
Unsubscribe:    [e-mail] "unsubscribe fractdev"


-------------------------------------------------------------------------------

From: "Lee H. Skinner" <[e-mail]>
Subject: Re: (fractdev) Autologmap bug
Date: 19 Aug 2001 04:12:00 -0600

>> Did you ever notice that the autologmap routine misses color #2?  This
problem goes back to the changes I did for calculating the logmap on the
fly
(version 19.2). <<

Yes.  When I do an autologmap, I always subtract 2 from the auto result
to get a better looking image!

Thanks for using Fractdev, The Fractint Developer's Discussion List
Post Message:   [e-mail]
Get Commands:   [e-mail] "help"
Administrator:  [e-mail]
Unsubscribe:    [e-mail] "unsubscribe fractdev"


-------------------------------------------------------------------------------

From: Tim Wegner <[e-mail]>
Subject: Re: (fractdev) Rotate and Skew
Date: 28 Aug 2001 21:18:54 -0600

A while ago Jonathan wrote:

> Take the following PAR, start it, zoom all the way out once.  Let the image
> generate, save it, and then make a PAR from it.  Run the PAR, set debug=50,
> and then reload the saved image.  You should see a pixel pattern that would
> indicate that one of the images is shifted.  Rotated at least.  This doesn't
> happen if you zoom in.  At least not until you get to the ap-math.
>
> Jonathan
>
> bob_and_weave      { ; ;
>                      ;  Jonathan Osuch [e-mail]
>                      ;  Aug 08, 2001 at 20:22:50
>                      ; Version 2001 Patchlevel 13
>   reset=2001 type=mandel
>   center-mag=-0.99428940396649250/+0.29970533391439380/1.276515e+011/1/-37\
>   .4999/-14.9313 params=0/0 float=y maxiter=1000 inside=0 sound=off
>   }

I tried this on Susan's laptop as I was flying over Iowa last
weekend. I couldn't quite see the problem. If you are still worried
about this, let me know. I am missing something in your
instructions.

Tim

Thanks for using Fractdev, The Fractint Developer's Discussion List
Post Message:   [e-mail]
Get Commands:   [e-mail] "help"
Administrator:  [e-mail]
Unsubscribe:    [e-mail] "unsubscribe fractdev"


-------------------------------------------------------------------------------

From: "Jonathan Osuch" <[e-mail]>
Subject: Re: (fractdev) Rotate and Skew
Date: 29 Aug 2001 19:26:08 -0500

Tim,

> I tried this on Susan's laptop as I was flying over Iowa last
> weekend. I couldn't quite see the problem. If you are still worried
> about this, let me know. I am missing something in your
> instructions.

The problem is that the PAR doesn't generate an exact duplicate of the
image.  The PAR image is rotated from the original.  I'll look at it closer
to see if I can tell what is happening.

BTW, with the fix to the logmap routine so that color #2 is displayed, I'll
need to increment the version number for backwards compatibility.

Jonathan



Thanks for using Fractdev, The Fractint Developer's Discussion List
Post Message:   [e-mail]
Get Commands:   [e-mail] "help"
Administrator:  [e-mail]
Unsubscribe:    [e-mail] "unsubscribe fractdev"


-------------------------------------------------------------------------------

From: "Jonathan Osuch" <[e-mail]>
Subject: Re: (fractdev) Rotate and Skew
Date: 29 Aug 2001 20:55:49 -0500

Tim,

> I am missing something in your instructions.

I've found the problem with the zoom-out.  I was missing a place where the
precision was getting set in the PAR.  But, even with the increased
precision, if you zoom in twice to get to the arbitrary precision
calculations the problem recurs.

Zoom in twice.  Make a GIF of the completed image.  Then make a PAR (you
could make the PAR at any time after the two zooms).  Run the PAR.  Compare
the image generated from the PAR to the image made immediately after zooming
in (debug=50).  You should see a bunch of pixels left on the screen because
the two images don't line up.

Jonathan



Thanks for using Fractdev, The Fractint Developer's Discussion List
Post Message:   [e-mail]
Get Commands:   [e-mail] "help"
Administrator:  [e-mail]
Unsubscribe:    [e-mail] "unsubscribe fractdev"