alharbin / dpix Goto Github PK

What steps will reproduce the problem?
1. open a model with a cylinder shape in it.

What is the expected output? What do you see instead?

Somehow the mitreing of the quad strips should change when the mitre angle 
is too high and causes the quad to become degenerate. Not sure exactly the 
solution here.

It would be nice to be able to visualize edges in the mesh.

when dpix is launched and no style is loaded then the default line texture
has a long inscrutable name, and the line color is bright red. probably
better would be to have a blank texture name and black lines.

What steps will reproduce the problem?
1. Set a 2d point of attention
2. Change window size
3.

What is the expected output? What do you see instead?

The focal point moves in a way that is inconsistent with the way the model 
itself moves. It would be 
best for the focal point to stay in the same place on the model as the window 
shifts.

Please use labels and text to provide additional information.

What steps will reproduce the problem?
1. trying opening a model on the command line with a relative path while 
not in libnpr

What is the expected output? What do you see instead?

The path created is wrong, it should be relative to the current working dir 
but it is relative to libnpr.

What steps will reproduce the problem?
1. Open a smooth model
2. Turn on flat shading
3.

What is the expected output? What do you see instead?

The model is still lit with smoothly varying color.

Several pen style features don't work in GPU mode yet. Among these are:

Endcaps
Texture blending
line density control (probably a separate issue)

Polygons have a basic PVS system. The same system could be applied to the 
lines. This would reduce the memory consumption of the segment atlas code.

What steps will reproduce the problem?
1. Load dpix with segment atlas on and a box
2. don't move the camera

What is the expected output? What do you see instead?

Line visibility is broken for the first frame. Moving the camera resets 
something and fixes the problem.

It appears that the GLSL code is not reading the value out of the depth 
texture properly. Channel "a" works but channel "b" does not.

What steps will reproduce the problem?
1. Load an out of date style file

What is the expected output? What do you see instead?

You should see a dialog, but instead it just prints a message to the console.

A good set of fly mode camera controls would be nice. Probably just wasd + 
some kind of lock to make you walk along a major axis plane while looking 
around.

What steps will reproduce the problem?
1. load box
2. switch to segment atlas

What is the expected output? What do you see instead?

On mac at least, the first frame after switching to the segment atlas will have 
zero returned from 
sumSegmentLengths. This may also account for why the lines disappear after 
capturing a 
screenshot on pc.

Please use labels and text to provide additional information.

What steps will reproduce the problem?
1. Set the bitrate to something huge in the settings dialog.
2. export a movie.
3.

What is the expected output? What do you see instead?

The exported movie does not appear to have the high bitrate.
.

The shader loader concatenates everything and then gives back line numbers 
based on the concatenation. However, it could keep track of the line 
numbers for each included file and provide a more helpful error message.

What steps will reproduce the problem?
1. Load an object such as the clevis
2. Zoom up to places such as the holes
3.

What is the expected output? What do you see instead?

Some of the lines seem to be checking their visibility in the wrong place. 
This seems to have happened since I made the change to compute the offsets 
in clip space rather than window space. 

Please use labels and text to provide additional information.

What is the expected output? What do you see instead?

warning level 1 should be "more warnings" than 0, but it is the reverse.

Some way to visualize and edit camera paths after creation would be helpful 
when creating long flythroughs. 

A cool feature could be to create a haloing effect around lines by using 
multiple colors in a stroke texture. The center color would be the pen 
color, and the outer color would be the background color. In this way lines 
drawn on top of other lines would appear to have a background colored halo.

What is the expected output? What do you see instead?

Sometimes sketchup models have flipped normals. The segment atlas code 
flips them back before loading into FBO. Maybe we should do the same with 
the spine test code somehow (though not clear how to do this).

Please use labels and text to provide additional information.

What steps will reproduce the problem?
1. load galleon
2. make a view such that the masts or sails are partially off screen.

What is the expected output? What do you see instead?

The quads should clip properly, but seem to clip a bit early (before the 
edge of the screen). Fiddling with the clip boundaries in clip_buffer.frag 
makes it go away, but shouldn't be necessary.

Please use labels and text to provide additional information.

What steps will reproduce the problem?
1. Save a dps file with the segment atlas turned on (galleon, my modern) 
2. load dpix from the command line with the dps, or as the first model loaded

What is the expected output? What do you see instead?

Sometimes it works fine. Sometimes it will get bogus values for the total 
number of samples 
(something like MAX_INT, etc.) , and the position of the lines will be totally 
corrupted.

What steps will reproduce the problem?
1. Open dpix
2. move some windows around
3. close, then open again

What is the expected output? What do you see instead?

Windows should stay in set places from last time, but do not.

[deleted issue]

A different blending function could be used for the invisible lines such that 
their color is affected by the polygons "in front".

What steps will reproduce the problem?
1. Save a dps file with the segment atlas turned on (galleon, my modern) 
2. load dpix from the command line with the dps, or as the first model loaded

What is the expected output? What do you see instead?

Sometimes it works fine. Sometimes it will get bogus values for the total 
number of samples 
(something like MAX_INT, etc.) , and the position of the lines will be totally 
corrupted.

Other settings may also make more sense as style parameters, for example, 
filtering priority or visibility.