Opened 13 years ago
Closed 13 years ago
#2188 closed defect (fixed)
read pixel pushes an int to stack
Reported by: | tonyforster | Owned by: | walter |
---|---|---|---|
Priority: | Unspecified by Maintainer | Milestone: | Unspecified |
Component: | Turtleart | Version: | Unspecified |
Severity: | Unspecified | Keywords: | dextrose |
Cc: | Distribution/OS: | Unspecified | |
Bug Status: | Unconfirmed |
Description
read pixel rounds colour values to an int, thats 0,0,0 to 1,1,1 or a colour space of only 8 colours.
Colour values are complements of what I would expect, ie 101 for green where i would expect 010, black 111 rather than 000
I dont understand turtle sees, print turtle sees always gives me 0
Attachments (6)
Change History (17)
comment:1 Changed 13 years ago by walter
comment:2 Changed 13 years ago by tonyforster
running on xo1, os373pyg, ta94
comment:3 Changed 13 years ago by tonyforster
- Keywords dextrose added
Changed 13 years ago by tonyforster
comment:4 Changed 13 years ago by tonyforster
more on print turtle sees
its not always 0, i just chanced a few 0's
set color turtle sees
red 0 30
orange 10 70
yellow 20 70
green 30 0
bu/gn 40 0
lt blu 50 0
mid blu60 0
dk blu 70 0
blu/vio80 30
vio 90 30
Changed 13 years ago by tonyforster
comment:5 Changed 13 years ago by walter
Bill,
Attached is a .ta file I used to test the Turtle Sees block as well as the output image from the test. My results are quite different from yours.
Please try using http://wiki.sugarlabs.org/images/3/37/TurtleArt-95.xo for your test.
thanks.
comment:6 Changed 13 years ago by walter
Also attached is the output of the read pixel test.
Note that since the color values in the palette are 8,8,8 bit rgb and the XO is 5,6,5, we'll not necessarily ever get an exact match. Currently I shift the red and blue values by 3 and the green values by 2. Perhaps I should change the palette to use the 5, 6, 5 colors so that we get an exact match (but this is only going to work for gray == 100 and shade == 5.
comment:7 Changed 13 years ago by tonyforster
V95 fixes it
Should let Bernie know V93 is in the pyg release candidate
Changed 13 years ago by tonyforster
comment:8 Changed 13 years ago by tonyforster
color != turtle sees
the error is quite big for shade !=50
havent tried gray
image attached
comment:9 Changed 13 years ago by tonyforster
uploaded screenshot has the png in filename mangled, logged as http://bugs.sugarlabs.org/ticket/1638#comment:3
comment:10 Changed 13 years ago by walter
I suppose that if shade or gray are not at 50, 100 respectfully, I could try to do a conversion before trying to map to the palette. (Originally I was going to ignore these cases as being inherently too noisy, but I'll test it.)
comment:11 Changed 13 years ago by walter
- Resolution set to fixed
- Status changed from new to closed
will be included in c95 as per http://git.sugarlabs.org/projects/turtleart/repos/mainline/commits/16d7d2a54607c38c4539a23dc266d359c49f82a6
Hmm. Not at all the expected behavior.
You are running on an XO 1 or XO 1.5?
I've tested extensively on XO 1 (362py) and it all works as expected. (Also in sugar-jhbuild.)
I'll come up with a test .ta file so we can have a common ground for debugging.