Hi all, please take the time to teach me what you know. I'm quite new to the world of encoding.
Attached below are 2 images from a video. 1st is an AVCHD (H264, 50hz, 25Mbps) 1080p recording from a Sony NEX camera.
2nd is re-encoded with Avidemux at mostly the same specs, except added "Asharp" Filter at [Threshold 2, Strength 10, Block Adaptive 4]. I even buffed the output bitrate to 30Mbps to cover for the added sharpen detail. All H264 settings are set to the best quality according to "www.avidemux.org/admWiki/doku.php?id=tutorial:h.264".
As is pretty obvious, there's too much noise in the image, especially near the edges. Look to the black railings top left, there is too much distortion along the railings to be the fault of "Asharp". There is none in the original image. The collars, sleaves, arms, pretty serious.
My main goal is to sharpen the video while maintaining smoothness. Msharpen seems too mild for this blur video. Please advice me on the best way to sharpen a blur video while still maintaining video quality, & not creating artifacts (even when outputting to higher bitrates!!). Ultimately, the video is to be uploaded on Youtube, & i can't upload a video like this as Youtube's compression will further pulverize it >.<
Quote from: dare2die on May 13, 2013, 10:39:39 PM
My main goal is to sharpen the video while maintaining smoothness.
I fear this is a contradiction, not a goal. I've never seen a video sharpened by a filter without more or less (of course normally less) effects as described by the examples you supplied.
Quote from: dare2die on May 13, 2013, 10:39:39 PMMsharpen seems too mild for this blur video.
I would guess the settings of msharpen a too strong.
Thanks for your reply. May i then rephrase that, my goal would be to sharpen the video while minimizing noise, creating unblocky *edges* while making facial features & textures stand out.
And the sharpen looks ok in the preview, it's only *AFTER* encoding that the image get's noisy. Many "encoding groups" on P2P somehow manage to get superb & detailed quality out of small filesizes, I wish i could learn their techniques.
Please don't stop there, give some advice on preferable settings from your experiences. Encoder settings, filter settings... for example, what would work best on a vid like the one i posted? Is it possible to make their faces clearer? Do i need to encode in an even higher bitrate? Are there settings in x264 that effect this? (I've already lowered Psychovisual RDO cause i read in wiki that it *add's* detail) Would more deblocking help or would it just ruin the details?
Thanks beforehand for all replies.
you did preview window to play end result before encoding:
Preview mode: (https://avidemux.org/smif/proxy.php?request=http%3A%2F%2Fwww.avidemux.org%2FadmWiki%2Flib%2Fexe%2Ffetch.php%3Fmedia%3Dusing%3Aoutput-qt.png&hash=490ba8a6d3adaf40c6255019b0d5cc148be0fe52) Output ââ,¬â€œ display filtered output in the main window.
source: Avidemux wiki (http://www.avidemux.org/admWiki/doku.php?id=using:main_window_-_qt_version#preview_mode)
Try with one minute section that causes the problem, until you get the result you want. I would first test the same clip you showed the pictures from. And re encode that one minute without the Asharp filter and see if there no artifacts occurring @ black railings top left.
Thanks for your reply, so your advice is that i keep testing just one minute with various settings to see what works. I'll do that.
Is there anyone who has personal favorite settings for this scenario who would like to share? (sorry i keep asking, but i wanna learn as much as i can from the veterans)