(Right click and open those images at original resolution to get the proper idea; Twitter always scales by default)
-
Show this thread
-
I would rate the artifacts at 720p as a bit worse than the LGP2, but not terrible (chinese left, LGP2 right). This is after adjusting the brightness/contrast/saturation/hue on both to attempt to match the original image (neither has good defaults).pic.twitter.com/sVYN4OVp5t
1 reply 0 retweets 8 likesShow this thread -
So I would call this a fairly capable 720p capture card overall, and only slightly lower performance at that resolution than devices that cost 10 times as much. At 1080p things are a bit messier, but it does work.
3 replies 0 retweets 22 likesShow this thread -
Replying to @marcan42
Any idea why they seem to actively have applied filtering that makes the quality significantly worse? I could understand if it was an analogue capture since the filtering is harder to get right but this is HDMI...
1 reply 0 retweets 0 likes -
Replying to @modwizcode
Maybe it can't actually encode 2 megapixels worth of JPEG, and they cheat by halving the resolution and just padding the coefficients with zeroes on the right side?
1 reply 0 retweets 1 like -
Replying to @marcan42
I totally overlooked the need to do the MJPEG compression. Still, it does seem to have some sort of capture ASIC and I feel like MJPEG would be an obvious built-in feature, and 2M is not... a large image. That probably wouldn't explain the weird sharpening for 720p though.
1 reply 0 retweets 0 likes -
Replying to @modwizcode
The MJPEG is definitely done in hardware, but that IP block might be underpowered. Or maybe it's a bitrate thing. Or a limitation of the pixel interface or some other layer.
2 replies 0 retweets 1 like -
Replying to @marcan42 @modwizcode
Are the captured frames the same or similar for p30 and p60 input ? An intermediate 960x1080p60 with half the pixel clock could explain the artifacts, but then I don't know how it could transform into 1080p30 without any DRAM.
1 reply 0 retweets 0 likes -
Replying to @LYhuel @modwizcode
It needs RAM to stream YUV frames slower anyway... And only having 1 megapixel worth of built-in RAM would explain the limitation too.
2 replies 0 retweets 1 like -
Replying to @marcan42 @modwizcode
It could be even less, an YUV frame might come from several input frames, with tearing artifacts. MJPEG only really needs 8 lines of image data, the space to hold the dct coefficients for those, then Huffman compression and USB buffers.
1 reply 0 retweets 0 likes
I don't think I saw any tearing in YUV mode, that'd be quite terrible :-)
Loading seems to be taking a while.
Twitter may be over capacity or experiencing a momentary hiccup. Try again or visit Twitter Status for more information.