Just letting people know I found a better way to capture your whole render
scene:
mapWidth, mapHeight = <the size of the scene you want to
capture>
mapTexture = pyglet.image.Texture.create(mapWidth, mapHeight,
internalformat=GL_RGB) # Force RGB because RGBA will mess up your blending.
glViewport(0, 0, mapWidth, mapHeight)
glMatrixMode(GL_PROJECTION)
glLoadIdentity()
glOrtho(0, mapWidth, 0, mapHeight, -1, 1)
glMatrixMode(GL_MODELVIEW)
glLoadIdentity()
id = GLuint(0)
glGenFramebuffersEXT(1, byref(id))
glBindFramebufferEXT(GL_FRAMEBUFFER_EXT, id)
glFramebufferTexture2DEXT(
GL_FRAMEBUFFER_EXT,
GL_COLOR_ATTACHMENT0_EXT,
mapTexture.target,
mapTexture.id,
mapTexture.level)
glClear(GL_COLOR_BUFFER_BIT)
glEnable(GL_BLEND)
glBlendFunc(GL_SRC_ALPHA, GL_ONE_MINUS_SRC_ALPHA)
<draw your entire scene here or whatever you want>
glBindFramebufferEXT(GL_FRAMEBUFFER_EXT, 0)
# Use overridden get_image_data so it can pull as RGB, not RGBA
get_image_data(mapTexture.owner).get_region(mapTexture.x,
mapTexture.y,
mapTexture.width,
mapTexture.height).save('full_screenshot.png')
# Reset your viewport back to normal
glViewport(0, 0, windowWidth, windowHeight)
glMatrixMode(GL_PROJECTION)
glLoadIdentity()
glOrtho(0, windowWidth, 0, windowHeight, -1, 1)
glMatrixMode(GL_MODELVIEW)
glLoadIdentity()
However upon doing this I discovered that blending did not work correctly
as everything was being blended with the invisible alpha, not the black
background you normally see.
I discovered alpha was still being written with the encoder despite the
texture being created as GL_RGB, and traced it back to the get_image_data.
It defaults all to RGBA. So you need to override the default get_image_data
to remove the alpha channel:
def get_image_data(texture, z=0):
'''Get the image data of this texture.
Changes to the returned instance will not be reflected in this
texture.
---- Overridden to format as RGB and GL_RGB instead of forced RGBA and
GL_RGBA
:Parameters:
`z` : int
For 3D textures, the image slice to retrieve.
:rtype: `ImageData`
'''
glBindTexture(texture.target, texture.id)
# Always extract complete RGBA data. Could check internalformat
# to only extract used channels. XXX
format = 'RGB'
gl_format = GL_RGB
glPushClientAttrib(GL_CLIENT_PIXEL_STORE_BIT)
glPixelStorei(GL_PACK_ALIGNMENT, 1)
buffer = \
(GLubyte * (texture.width * texture.height * texture.images *
len(format)))()
glGetTexImage(texture.target, texture.level,
gl_format, GL_UNSIGNED_BYTE, buffer)
glPopClientAttrib()
data = ImageData(texture.width, texture.height, format, buffer)
if texture.images > 1:
data = data.get_region(0, z * texture.height, texture.width,
texture.height)
return data
After I did that everything worked perfectly.
On Sunday, November 12, 2017 at 2:31:36 PM UTC-6, Charlie wrote:
>
> This method worked, thanks!
>
> On a side note, I tried to do multiple viewports using glViewport, but for
> some reason the whole window defaults to the second viewport even if I set
> another viewport right after. Not sure what I am missing on that one.
>
> On Wednesday, November 1, 2017 at 3:44:21 AM UTC-5, Benjamin Moran wrote:
>>
>> I can think of a hacky way to do it that might work, but would be a bit
>> wasteful.
>> Maybe you can save the current window buffer, and then use gltranslate to
>> move the view, and take another screenshot.
>> All of these images could then be saved to one big texture, which you can
>> save out as a single png.
>> You would need to force a screen redraw in between each translate.
>>
>> I have no idea if that would work, but maybe!
>>
>>
>>
>> On Wednesday, November 1, 2017 at 2:16:32 AM UTC+9, Charles wrote:
>>>
>>> Yeah I figured that was the case with the culling.
>>>
>>> I am trying to get a screenshot with a 1:1 ratio, kind of like a level
>>> editor can save a shot, but without zoom so I don't lose quality and
>>> detail. I've tried messing with viewport and projection settings, but seems
>>> the GL buffer still can't go outside that. The screenshot produced shows
>>> the correct dimensions of the viewport, but only the window part has
>>> anything on it, the rest ends up black.
>>>
>>> On Sunday, October 29, 2017 at 4:59:18 AM UTC-5, Benjamin Moran wrote:
>>>>
>>>> Hi Charles,
>>>>
>>>> I don't think there are any GL buffers that would normally have this
>>>> image data, because the GPU would cull this off-screen data from the final
>>>> fragment rasterization.
>>>>
>>>> Are you trying to just capture the data to save a screen shot, or do a
>>>> mini map type effect?
>>>>
>>>> I would guess that most level editors simply render the scene twice.
>>>> The main view, and a second view with different viewport and projection
>>>> settings. That would be zoomed out in this case.
>>>>
>>>>
>>
--
You received this message because you are subscribed to the Google Groups
"pyglet-users" group.
To unsubscribe from this group and stop receiving emails from it, send an email
to [email protected].
To post to this group, send email to [email protected].
Visit this group at https://groups.google.com/group/pyglet-users.
For more options, visit https://groups.google.com/d/optout.
