Screen recording is not very clear for characters

[George2]

Hello everyone,

I am using Windows Media Encoder 9 on Windows Vista. And I find the recorded video quality is not very good, the major issue is the details are not clear enough -- for example, character is not very clearly displayed, and another example is when list a folder in explorer containing a lot of files (file names are not very clear), when quickly scroll-up/scroll-down a folder, the file names are not displayed immediately, but displayed gradually. I am using full screen recording (CBR) mode.

Here is my recorded video. Any advice how to improve quality?

http://tinypic.com/player.php?v=jrp73s&s=5

thanks in advance,
George

[tomlee59]

I think your expectations may be too high. You're not going to get computer-screen resolution on these things -- the file size would be too great. To shrink the files, many compromises are made, and resolution is certainly one of them.

The best you can do is to create the highest resolution, highest bitrate sourcefile that the hosting site (or whatever the end destination is) will accept.

[poisondeathray]

I'm assuming that was just an example for display/discussion purposes (which was flash, not wmv), and not the actual video

What was the bitrate and fps?

The delay lag when scrolling is probably from a low fps

You could try using a higher bitrate and fps, or recording to a lossless format then editing (with all your PIP overlays etc...) and encoding using VBR

Or, you will definitely get better quality using h.264 (you can use lower bitrates than wmv, and still the characters will be more crisp compared to wmv)

[George2]

Hi tomlee59,

I suspect there is some configuration error. Because when I use Screen Capture mode of Windows Media Encoder, the characters are very clear even if I only use 300Kb bit rate. Currently I am using Full Screen Recording mode of Windows Media Encoder, and even if I use 1000Kb bit rate, the character is not clear. Any ideas what is wrong?

BTW: the reason why I do not use Screen Capture mode is, it does not record backgroud of desktop quite well. Full Screen mode could record desktop background quite well, but characters not very clear.

I think your expectations may be too high. You're not going to get computer-screen resolution on these things -- the file size would be too great. To shrink the files, many compromises are made, and resolution is certainly one of them.

The best you can do is to create the highest resolution, highest bitrate sourcefile that the hosting site (or whatever the end destination is) will accept.

[George2]

Thanks poisondeathray, the bit rate is about 1000Kb and fps is about 15. I have tried to use fps to 30, still the same effect.

1.

What means "an example for display/discussion purposes (which was flash, not wmv), and not the actual video"? I am confused about what do you mean example for ... display/discussion purpose ... not the actual video. Could you say in some other words please?

2.

The reason why I do not use Screen Capture mode is, it does not record backgroud of desktop quite well. Full Screen mode could record desktop background quite well, but characters not very clear. Any ideas what is wrong?

3.

I think Windows Media Encoder does not support h.264 correct?

I'm assuming that was just an example for display/discussion purposes (which was flash, not wmv), and not the actual video

What was the bitrate and fps?

The delay lag when scrolling is probably from a low fps

You could try using a higher bitrate and fps, or recording to a lossless format then editing (with all your PIP overlays etc...) and encoding using VBR

Or, you will definitely get better quality using h.264 (you can use lower bitrates than wmv, and still the characters will be more crisp compared to wmv)

[poisondeathray]

Thanks poisondeathray, the bit rate is about 1000Kb and fps is about 15. I have tried to use fps to 30, still the same effect.

Are you dropping frames, or getting the full 30fps? You reported lagging in another thread when opening up powerpoint....so it's entirely possible you are dropping frames. 15fps explains the slow scroll movement and lag/delay, but not the poor image quality. At 1000kbps it should be fine, even with wmv

What means "an example for display/discussion purposes (which was flash, not wmv), and not the actual video"? I am confused about what do you mean example for ... display/discussion purpose ... not the actual video. Could you say in some other words please?

I meant the link you provided is a re-encoded copy of your original, so it's not entirely representative of what you are trying to say. It would have been better to upload the original copy to a free hosting site (e.g. mediafire.com, megaupload.com, or even here if <5MB)

I think Windows Media Encoder does not support h.264 correct?

Correct.

Post the exact settings you used to record with these poor results. The details are important

[tomlee59]

Hi tomlee59,

I suspect there is some configuration error. Because when I use Screen Capture mode of Windows Media Encoder, the characters are very clear even if I only use 300Kb bit rate. Currently I am using Full Screen Recording mode of Windows Media Encoder, and even if I use 1000Kb bit rate, the character is not clear. Any ideas what is wrong?

Please tell us exactly what you are doing, what the settings are, etc. As I mentioned above, for best results, resolution and bitrate matter. Just increasing one of those alone may not be sufficient.

[George2]

Thanks, tomlee59!

Here is all of my settings, let me know if you need to know more.

http://i42.tinypic.com/fbgupv.jpg

http://i42.tinypic.com/2a7g5so.jpg

http://i42.tinypic.com/2qlqr78.jpg

Here is my Windows Media Encoder configuraiton file => http://www.mediafire.com/?jnycjnl5qvt Could you reproduce my issue?

Hi tomlee59,

I suspect there is some configuration error. Because when I use Screen Capture mode of Windows Media Encoder, the characters are very clear even if I only use 300Kb bit rate. Currently I am using Full Screen Recording mode of Windows Media Encoder, and even if I use 1000Kb bit rate, the character is not clear. Any ideas what is wrong?

Please tell us exactly what you are doing, what the settings are, etc. As I mentioned above, for best results, resolution and bitrate matter. Just increasing one of those alone may not be sufficient.

[George2]

Thanks for all of your great ideas, poisondeathray!

1.

Here is my original file and all of my settings, let me know if you need more information or have any good ideas?

http://www.mediafire.com/?tzjccxdowny

http://i42.tinypic.com/fbgupv.jpg

http://i42.tinypic.com/2a7g5so.jpg

http://i42.tinypic.com/2qlqr78.jpg

Here is my Windows Media Encoder configuraiton file => http://www.mediafire.com/?jnycjnl5qvt Could you reproduce my issue?

2.

"Are you dropping frames, or getting the full 30fps?" -- how could I check this?

Thanks poisondeathray, the bit rate is about 1000Kb and fps is about 15. I have tried to use fps to 30, still the same effect.

Are you dropping frames, or getting the full 30fps? You reported lagging in another thread when opening up powerpoint....so it's entirely possible you are dropping frames. 15fps explains the slow scroll movement and lag/delay, but not the poor image quality. At 1000kbps it should be fine, even with wmv

What means "an example for display/discussion purposes (which was flash, not wmv), and not the actual video"? I am confused about what do you mean example for ... display/discussion purpose ... not the actual video. Could you say in some other words please?

I meant the link you provided is a re-encoded copy of your original, so it's not entirely representative of what you are trying to say. It would have been better to upload the original copy to a free hosting site (e.g. mediafire.com, megaupload.com, or even here if <5MB)

I think Windows Media Encoder does not support h.264 correct?

Correct.

Post the exact settings you used to record with these poor results. The details are important

[jagabo]

You are taking whatever your full screen resolution is (1024x768?) and reducing it to 640x480 for recording. That's the source of your problem.

[George2]

Hi jagabo, I did not reduce it to 640x480, how do you find such information?

You are taking whatever your full screen resolution is (1024x768?) and reducing it to 640x480 for recording. That's the source of your problem.

[jagabo]

Look at the frame size of your WMV file -- 640x480.

[George2]

Thanks jagabo, you are correct! I have tried again, using the same configuration file with 1024x768 output (i.e. the same as input resolution), the quality is still not good -- for characters. Here is the the recorded media file and my Windows Media Encoder configuration file. Do you have any ideas what is wrong?

(from the recored video, you can see characters are not clear, showing gradually.)

media file:

http://www.mediafire.com/?jzdblnjldzz

Windows Media Encoder configuration file

http://www.mediafire.com/?5gfwnzg4zdr

Look at the frame size of your WMV file -- 640x480.

[jagabo]

Your bitrate is now too low for the resolution.

The higher the resolution -- the more bitrate you need.
The more detail in each frame -- the more bitrate you need.
The more the picture changes from frame to to frame -- the more bitrate you need.
The higher the frame rate -- the higher the bitrate you need.
The higher the contrast -- the higher the bitrate you need.

[poisondeathray]

Also, your current settings are 500kbps, 5fps, 50 image quality (sharpness)

I would bump up the fps to >20 to 30 , 100 image quality, increase the bitrate to 750 to 1000, and use WMV 9 Screen instead of WMV 9. When you enter your changes, press OK then apply.

The gradual showing of characters (delay) is partly due to the low fps, but also a CBR capped encode means the encoder doesn't have the ability to go above that limit. So when you have a quick movement or a "scene change", the quality will be worse compared to a VBR encode, which will allocate the bitrate more properly (more to areas where it is needed, and less to areas where it is not needed); thus a VBR encode is more efficient. Of course 2pass VBR is not possible for capturing directly.

If you plan on editing or adding effects / overlays / PIP that sort of thing after, that this is not a good capture format. It is lossy and when you re-encode to apply those effects, the quality will drop again - hence the recommendation to capture to a lossless format, then edit & encode to the final distribution format. Even if you plan on sticking with WMV, you get better efficiency/quality by using VBR (and possibly VC-1) encoding from the lossless format. At the very least, you should use a high bitrate >2000kbps CBR capture if you plan on farther editing,

[Bjs]

Like squeezing an elephant onto a postage stamp ... fonts will never be clear enough to satisfy.

For dvd output you might try lowering screen resolution down to 800x600, then final output as pal or nstc dv avi and re-encode.

Windows movie maker or what ever vista comes with should be able to convert and output as dv avi ... been a while since I used windows media encoder.

Other options would be to record only a selected region where it follows the mouse ... just a few idea's which I use often.

[George2]

jagabo, I have set bit rate to 5000K and still fonts are not clear. Could you reproduce?

I think besides bit rate, there should be some other settings to tune to make character more clearer. Any ideas?

Your bitrate is now too low for the resolution.

The higher the resolution -- the more bitrate you need.
The more detail in each frame -- the more bitrate you need.
The more the picture changes from frame to to frame -- the more bitrate you need.
The higher the frame rate -- the higher the bitrate you need.
The higher the contrast -- the higher the bitrate you need.

[George2]

Thanks poisondeathray, I have tested VBR could have better quality. For VBR, I have three questions,

1. I did not find where to set VC-1 profile?
2. For CBR, we could set bit rate (in the Compression Tab of properties view of recording session), but for VBR, I did not find a place to set bit rate. How to set bit rate for VBR?
3. There are 3 types of VBR, Quality VBR, bit rate VBR and bit rate VBR (peak), what are the differences between them and which one is most suitable for my scenario of screen recording?

BTW: Currently I am using quality VBR and it works much better and displays characters much clearer.

Another question, when we set 50, it should be image sharpness as dialog label indicates, why do you say "50 image quality (sharpness)" -- seesm sharpness is the same as quality?

Also, your current settings are 500kbps, 5fps, 50 image quality (sharpness)

I would bump up the fps to >20 to 30 , 100 image quality, increase the bitrate to 750 to 1000, and use WMV 9 Screen instead of WMV 9. When you enter your changes, press OK then apply.

The gradual showing of characters (delay) is partly due to the low fps, but also a CBR capped encode means the encoder doesn't have the ability to go above that limit. So when you have a quick movement or a "scene change", the quality will be worse compared to a VBR encode, which will allocate the bitrate more properly (more to areas where it is needed, and less to areas where it is not needed); thus a VBR encode is more efficient. Of course 2pass VBR is not possible for capturing directly.

If you plan on editing or adding effects / overlays / PIP that sort of thing after, that this is not a good capture format. It is lossy and when you re-encode to apply those effects, the quality will drop again - hence the recommendation to capture to a lossless format, then edit & encode to the final distribution format. Even if you plan on sticking with WMV, you get better efficiency/quality by using VBR (and possibly VC-1) encoding from the lossless format. At the very least, you should use a high bitrate >2000kbps CBR capture if you plan on farther editing,

[George2]

1. Could Windows Movie Maker record screen? Confused why you mention Windows Movie Maker here?

2. "Like squeezing an elephant onto a postage stamp ... fonts will never be clear enough to satisfy." -- confused. My input resolution is the same as output frame size, why you think it is resized?

3. "then final output as pal or nstc dv avi and re-encode." -- I think Windows Media Encoder only supports output as wmv, not avi. Please correct me if I am wrong.

Like squeezing an elephant onto a postage stamp ... fonts will never be clear enough to satisfy.

For dvd output you might try lowering screen resolution down to 800x600, then final output as pal or nstc dv avi and re-encode.

Windows movie maker or what ever vista comes with should be able to convert and output as dv avi ... been a while since I used windows media encoder.

Other options would be to record only a selected region where it follows the mouse ... just a few idea's which I use often.

[jagabo]

I don't use WME or the WMV codecs but I suspect there may be Min and Max bitrate settings along with the average bitrate setting. You may need to turn the Max bitrate setting higher.

[poisondeathray]

1. I did not find where to set VC-1 profile?
2. For CBR, we could set bit rate (in the Compression Tab of properties view of recording session), but for VBR, I did not find a place to set bit rate. How to set bit rate for VBR?
3. There are 3 types of VBR, Quality VBR, bit rate VBR and bit rate VBR (peak), what are the differences between them and which one is most suitable for my scenario of screen recording?

BTW: Currently I am using quality VBR and it works much better and displays characters much clearer.

Another question, when we set 50, it should be image sharpness as dialog label indicates, why do you say "50 image quality (sharpness)" -- seesm sharpness is the same as quality?

1) I don't think you can use the higher quality VC-1 for screen recording, it's meant for offline, not real time screen recording (very slow, you would drop frames). You can record to a lossless format, then to VC-1

2) Quality mode doesn't have a set bitrate, it depends on the source complexity and the quality setting. It might be big, it might be small.

3) Quality mode is just that: the quality level determines how much bitrate the encoder uses. It might be 5000kbps, it might be 1000kbps, or any number depending on source complexity. Your file might be huge , or small. Bit rate peak is a constrained profile. It's used for capping and streams that aren't allowed peaks (e.g. your bandwidth provider might have caps for max peaks). The quality is lower, because on scenes/frames that require large jumps (complex scenes), the quality drops because the bitrate isn't allowed to be high

Most people would use a lossless encode (eg. using camstudio (free) or camtasia), then encode to WMV offline. This allows the best quality and control. The overhead is much lower than encoding to WMV, and there are fewer sync issues (like dropped frames, you can use a higher fps without fear), and fewer editing issues. (Actually most people wouldn't use WMV, they would use h.264/aac for better quality AND smaller filesize).

If you were using this as the final format without farther editing, then most would use CBR. CBR allows for real time encoding with minimal dropped frames for screen recording, but is highly inefficient (frames that require low bitrate get too much ie. wasted bitrate, and frames that require big bitrate get too little). 1-pass VBR is better, but is slightly more CPU intensive and you can get dropped frames depending on your hardware. Obviously, you can't use the higher quality 2-pass VBR encode for real time. The 1pass VBR is not as good, because an analysis pass isn't run for bitrate allocation, so it is not as efficient either

4) "video smoothness" only refers to bitrate mode (not quality mode, which would use a quality level value). It ranges from 0-100, higher values give sharper results, at the expense of more cpu usage (possibly dropped frames depending on your hardware specs)

[George2]

Great reply, poisondeathray!

A few more comments/questions,

1. what is the differences between bit rate VBR and bit rate VBR (peak)?
2. In bit rate mode, what exactly does sharpness mean for end user? More sharper means better quality always?
3. what means "source complexity"? Could we control and define "source complexity" or it is automatically detected by encoder?
4. "overhead is much lower than encoding to WMV" -- I think you mean overhead of CPU used to encode while recording, and if recording the raw data (lossless mode) without encoding, CPU could have more bandwidth. My understanding correct?

1. I did not find where to set VC-1 profile?
2. For CBR, we could set bit rate (in the Compression Tab of properties view of recording session), but for VBR, I did not find a place to set bit rate. How to set bit rate for VBR?
3. There are 3 types of VBR, Quality VBR, bit rate VBR and bit rate VBR (peak), what are the differences between them and which one is most suitable for my scenario of screen recording?

BTW: Currently I am using quality VBR and it works much better and displays characters much clearer.

Another question, when we set 50, it should be image sharpness as dialog label indicates, why do you say "50 image quality (sharpness)" -- seesm sharpness is the same as quality?

1) I don't think you can use the higher quality VC-1 for screen recording, it's meant for offline, not real time screen recording (very slow, you would drop frames). You can record to a lossless format, then to VC-1

2) Quality mode doesn't have a set bitrate, it depends on the source complexity and the quality setting. It might be big, it might be small.

3) Quality mode is just that: the quality level determines how much bitrate the encoder uses. It might be 5000kbps, it might be 1000kbps, or any number depending on source complexity. Your file might be huge , or small. Bit rate peak is a constrained profile. It's used for capping and streams that aren't allowed peaks (e.g. your bandwidth provider might have caps for max peaks). The quality is lower, because on scenes/frames that require large jumps (complex scenes), the quality drops because the bitrate isn't allowed to be high

Most people would use a lossless encode (eg. using camstudio (free) or camtasia), then encode to WMV offline. This allows the best quality and control. The overhead is much lower than encoding to WMV, and there are fewer sync issues (like dropped frames, you can use a higher fps without fear), and fewer editing issues. (Actually most people wouldn't use WMV, they would use h.264/aac for better quality AND smaller filesize).

If you were using this as the final format without farther editing, then most would use CBR. CBR allows for real time encoding with minimal dropped frames for screen recording, but is highly inefficient (frames that require low bitrate get too much ie. wasted bitrate, and frames that require big bitrate get too little). 1-pass VBR is better, but is slightly more CPU intensive and you can get dropped frames depending on your hardware. Obviously, you can't use the higher quality 2-pass VBR encode for real time. The 1pass VBR is not as good, because an analysis pass isn't run for bitrate allocation, so it is not as efficient either

4) "video smoothness" only refers to bitrate mode (not quality mode, which would use a quality level value). It ranges from 0-100, higher values give sharper results, at the expense of more cpu usage (possibly dropped frames depending on your hardware specs)

[jagabo]

2. In bit rate mode, what exactly does sharpness mean for end user? More sharper means better quality always?

No. It takes more bitrate to encode sharp video than it does blurry video. If your source is very sharp and you encode with a low bitrate, maintaining all the sharpness will cause an increase in macroblock artifacts. So you'll get sharper video with more macroblocks.

3. what means "source complexity"? Could we control and define "source complexity" or it is automatically detected by encoder?

He's talking about the properties of the source video. A noiseless, black video is low complexity. A highly detailed, high action video is high complexity. The more complex the video, the higher the bitrate will turn out when using quality based encoding. Encoding is a two sided coin: You can specify the bitrate (size) and accept whatever quality you get, or you can specify the quality and accept whatever bitrate (size) you get.

4. "overhead is much lower than encoding to WMV" -- I think you mean overhead of CPU used to encode while recording, and if recording the raw data (lossless mode) without encoding, CPU could have more bandwidth. My understanding correct?

I'm not sure what you're saying but what poisondeathray meant is that encoding with a lossless codec while capturing can be less CPU intensive than encoding with WMV while capturing. This helps assure you don't drop frames while capturing. When capturing NTSC video a new frame arrives from the capture device every 1/30 of a second. The CPU must compress and save that frame before the next frame arrives. In order to do this WMV will have to take a lot of shortcuts. It will not be able the do the best possible job. If you capture with a lossless codec first then convert to WMV later, WMV can spend as much time as necessary to do a good job of compressing the video.

[poisondeathray]

1. what is the differences between bit rate VBR and bit rate VBR (peak)?

When selecting a 1-pass encode, it's the same thing as "average bitrate", so each frame gets higher or lower distribution within that average setting. The VBR (Peak) is also called "constrained" which means it caps peaks to that setting. So if you used a peak of 1000kbps for example, the encoder would impose a maximum limit. This is generally used only for streaming where ISP's place a limitation on peaks, or your are trying to load balance a server.

This potentially gives worse quality, because if the source complexity requires a larger bitrate (e.g. a big explosion or action sequence), the sequence will look very bad because of the cap. Also scene changes often place a new I-frame with a high bitrate, and following frames often reference that high quality I-frame. If that I-frame is capped, the following frames will be of lower quality as well

If I haven't pestered you enough about using h.264 (all the professional tutorials use this format), here is an example of 3 videos in a zip file. It is a short 14s example, 24fps, 960x540, similar to the one you uploaded (except smaller dimensions). Both used 2pass settings. The h.264 example is 350kb/s and has very clear legible type, with only minor artifacting ; the WMV example is almost twice the bitrate (2x the size) at ~675kb/s, yet still retains worse artifacting especially on motion (scrolling). The WMV at 385kb/s is completly illegible on the scrolling segments. If you were to restrict yourself to a 1pass VBR encode (instead of a 2pass VBR), your results would be even worse than this example.

The efficiency just isn't there using your current method and you will be compensating by using a much larger bitrate. Furthermore, if you were to process it later with PIP effects, overlays, etc... as you had planned in a previous thread, the quality will only deteriorate! This is a lossy format and the compression artifacts dramatically get worse as you re-encode. You are well on your way to a blurry mess, and I would hate to see that happen to educational tutorials



test1.zip

[George2]

Thanks jagabo!

1.

"No. It takes more bitrate to encode sharp video than it does blurry video." -- I am still confused about the term sharpness, is sharpness determined purely by the source property or determined by both source and recorder's expectation?

2.

My understanding is more sharpness means more clear details, and better quality. Not sure whether my understanding is correct?

3.

I am interested in what do you mean "macroblock artifacts"? Could you describe more about what it means please?

2. In bit rate mode, what exactly does sharpness mean for end user? More sharper means better quality always?

No. It takes more bitrate to encode sharp video than it does blurry video. If your source is very sharp and you encode with a low bitrate, maintaining all the sharpness will cause an increase in macroblock artifacts. So you'll get sharper video with more macroblocks.

3. what means "source complexity"? Could we control and define "source complexity" or it is automatically detected by encoder?

He's talking about the properties of the source video. A noiseless, black video is low complexity. A highly detailed, high action video is high complexity. The more complex the video, the higher the bitrate will turn out when using quality based encoding. Encoding is a two sided coin: You can specify the bitrate (size) and accept whatever quality you get, or you can specify the quality and accept whatever bitrate (size) you get.

4. "overhead is much lower than encoding to WMV" -- I think you mean overhead of CPU used to encode while recording, and if recording the raw data (lossless mode) without encoding, CPU could have more bandwidth. My understanding correct?

I'm not sure what you're saying but what poisondeathray meant is that encoding with a lossless codec while capturing can be less CPU intensive than encoding with WMV while capturing. This helps assure you don't drop frames while capturing. When capturing NTSC video a new frame arrives from the capture device every 1/30 of a second. The CPU must compress and save that frame before the next frame arrives. In order to do this WMV will have to take a lot of shortcuts. It will not be able the do the best possible job. If you capture with a lossless codec first then convert to WMV later, WMV can spend as much time as necessary to do a good job of compressing the video.

[George2]

Thanks poisondeathray!

1.

Could I understand in this way, bit rate VBR and bit rate VBR (peak) are both 1-pass VBR encoding, the only differences is bit rate VBR (peak) has a max bit rate limitation, but bit rate VBR has no such limitation? And both bit rate VBR and bit rate VBR (peak) results in expected average bit rate in result video?

2.

"because of the cap" -- I think cap you mean max bite rate? I am interested why max bit rate is called cap? Cap in English is hat, it is hard for me to understand that is means max of xxx.

3.

"as you had planned in a previous thread, the quality will only deteriorate! This is a lossy format and the compression artifacts dramatically get worse as you re-encode." -- Do you mean both h.264 and wmv are bad? Or only wmv is bad?

1. what is the differences between bit rate VBR and bit rate VBR (peak)?

When selecting a 1-pass encode, it's the same thing as "average bitrate", so each frame gets higher or lower distribution within that average setting. The VBR (Peak) is also called "constrained" which means it caps peaks to that setting. So if you used a peak of 1000kbps for example, the encoder would impose a maximum limit. This is generally used only for streaming where ISP's place a limitation on peaks, or your are trying to load balance a server.

This potentially gives worse quality, because if the source complexity requires a larger bitrate (e.g. a big explosion or action sequence), the sequence will look very bad because of the cap. Also scene changes often place a new I-frame with a high bitrate, and following frames often reference that high quality I-frame. If that I-frame is capped, the following frames will be of lower quality as well

If I haven't pestered you enough about using h.264 (all the professional tutorials use this format), here is an example of 3 videos in a zip file. It is a short 14s example, 24fps, 960x540, similar to the one you uploaded (except smaller dimensions). Both used 2pass settings. The h.264 example is 350kb/s and has very clear legible type, with only minor artifacting ; the WMV example is almost twice the bitrate (2x the size) at ~675kb/s, yet still retains worse artifacting especially on motion (scrolling). The WMV at 385kb/s is completly illegible on the scrolling segments. If you were to restrict yourself to a 1pass VBR encode (instead of a 2pass VBR), your results would be even worse than this example.

The efficiency just isn't there using your current method and you will be compensating by using a much larger bitrate. Furthermore, if you were to process it later with PIP effects, overlays, etc... as you had planned in a previous thread, the quality will only deteriorate! This is a lossy format and the compression artifacts dramatically get worse as you re-encode. You are well on your way to a blurry mess, and I would hate to see that happen to educational tutorials



test1.zip

[poisondeathray]

1. Could I understand in this way, bit rate VBR and bit rate VBR (peak) are both 1-pass VBR encoding, the only differences is bit rate VBR (peak) has a max bit rate limitation, but bit rate VBR has no such limitation? And both bit rate VBR and bit rate VBR (peak) results in expected average bit rate in result video?

Yes, the only difference between vbr and vbr (peak) is a cap. When you analyze a video, each frame requires a certain amount of bitrate in proportion to complexity. Below is an unrelated example for demonstration purposes:



You might constrain the peaks to a certain value, so everything above that value is cut off. This would push down the graph and distribute the allocated bitrate in a more flat distribution, which usually means lower quality. The average bitrate of the entire video is the same, the distribution will be different

But, both modes can apply to either 1 or 2 pass (there is a 2pass checkbox in the compression tab if you are using WME), and the average bitrate is the value you input. If you are screen recording directly to your final format (instead of doing a lossless recording then encoding to the final format), then you can't take advantage of the more accurate 2pass mode. 1 pass mode is also known as average bitrate mode (ABR) which is an inaccurate method that roughly estimates the value (absence of analysis pass), it can't as accurately examine the content compared to a 2pass mode, so the "valleys" and "peaks" aren't as pronounced. The 2pass mode looks at the entire video and allocates approriately, but the 1pass cannot, it just looks locally at the few frames and "guesses" an approximation within that average range. Since the reference frames are not as high quality, the entire video is worse (because the subsequent frames that are based on the reference frames "see" a lower quality version, which compounds the quality loss), and bitrate is "wasted" as certain frames get too much allocation, and other frames get too little - hence inefficient

2. "because of the cap" -- I think cap you mean max bite rate? I am interested why max bit rate is called cap? Cap in English is hat, it is hard for me to understand that is means max of xxx.

By "cap" I mean maximum limitation imposed. Sorry for the confusing terminology. So the peaks would be cut off if they exceeded that value. In the 2nd configuration tab, it's labelled "video peak bit rate"

3. "as you had planned in a previous thread, the quality will only deteriorate! This is a lossy format and the compression artifacts dramatically get worse as you re-encode." -- Do you mean both h.264 and wmv are bad? Or only wmv is bad?

They are both lossy and "bad" for capture purposes, but at the same bitrate (filesize) h.264 is much better, especially at low bitrates. You can look at the 350kbps example I provided - h.264 example is clearly legible, the wmv is not. Below is a cropped representative sample frame. It would take ~2x the size at this average bitrate target range for wmv to equal h.264 quality., because wmv3 is less efficient and lacks some advanced features that h.264 has like deblocking. Not only that, encoding to wmv is much slower, (partly because the multithreading support is poor for wmv, partly because it's just slooow). So on this example, the h.264 encode was about 3-4x faster on a quad core.

h.264 ~350kbps


wmv ~385kbps


If you have a crappy lossy source, then re-encode it with lossy compression, it gets worse. This applies to everything and a fact of life. You can test this out yourself. Just re-encode the same video over and over, and the compression artifacts exponentially worsen each time. It's just that the wmv gets worse a lot faster given the same bitrate (it's not as efficient). This is why you should be using a lossless screen recording format (unless you aren't going to edit or add titles, overlays, PIP that sort of thing), because you minimize generation losses. Remember these are final distribution formats, they are not meant for editing, and they certainly are not edit friendly. If you insist on using a lossy capture format, at the very least use a very high bitrate to minimize quality loss for subsequent stages.

[George2]

Thanks poisondeathray,

h.264 is great, the issue which blocks me from using h.264 is since all of my legacy applications are developed on .Net platform, I do not find a good SDK (like Microsoft Expression Encoder 2 SDK or Microsoft Media Encoder SDK) which integrates codec with .Net quite well. Another issue is, for h.264 encoder, I have to let each end user install h.264 codec -- a little inconvenient for end user, but for wmv format, Windows has built-in codec.

Any comments or advice for my two issues?

have a good weekend,
George

[poisondeathray]

My programming knowledge is limited, but for example, megui (front end for the open source x264 encoder) is based on .net framework, so you might be able to use a similar model

If we assume your end users are low on computer literacy, what was your proposed workflow for them and apps that you intended them to use/install?

Won't they have to install other programs such as editing programs, or your recording script? What are you expecting them to do? Just record the screen and keep it as is (no editing or post processing) ? Is this meant for local playback or will there be an internet component (will some recordings be hosting online, for example)?

For screen recording, most people use the free camstudio, or camtasia

If you record in any AVI VFW format, then you could use vdub, avidemux or any number of free open source programs to edit. You could even use WAX to do your PIP overlay. If you recall there are issues with WMV due to licensing and it's proprietary nature.

For encoding, there is a free h.264 encoder (x264). It must be installed separately, but there are many easy to use front ends, eg. ripbot264, handbrake, xvid4psp.

For decoding , there are several self contained players e.g. vlc, kmplayer, smplayer, which contain all the necessary codecs for hundreds of video formats. I think VLC even has a "portable" installation option, so there is no system install, it can come in a "usb" format.

A multipurpose "swiss army knife" codec suite is ffdshow, which contains many options for decoding and encoding, it's probably the most commonly installed codec.

If you want to keep it as simple as possible, you could bundle the apps or possibly write one yourself (all the software for my proposed method is free, most open source). WMV is certainly an option, but you need to instruct the end users to record at higher bitrates and fps, and most your quality issues will improve. Editing like PIP, overlays is difficult but can be done (especially with avisynth, but it's probably much to difficult for your users). Another factor for consideration is what kind of hardware do they have? WMV is very CPU resource hungry and you will be in danger of dropping frames (unless you use low quality, low fps or tiny screen dimension); this is because you are encoding (compressing) audio & video on the fly. But even old computers can capture with "proper" capture codecs without dropping frames.

[George2]

Thanks poisondeathray,

1.

I have tried megui, but I can not find from its main GUI that it supports screen recording (e.g. I can select to record screen as input source), let me know if I am wrong.

2.

Camtasia is free? I am using it and I remember we need to buy license key or else it is only trial version.

3.

What is the benefit of using AVI VFW format compared of using h.264, and also compared of using my current solution of using Windows Media Encoder to record to wmv?

My programming knowledge is limited, but for example, megui (front end for the open source x264 encoder) is based on .net framework, so you might be able to use a similar model

If we assume your end users are low on computer literacy, what was your proposed workflow for them and apps that you intended them to use/install?

Won't they have to install other programs such as editing programs, or your recording script? What are you expecting them to do? Just record the screen and keep it as is (no editing or post processing) ? Is this meant for local playback or will there be an internet component (will some recordings be hosting online, for example)?

For screen recording, most people use the free camstudio, or camtasia

If you record in any AVI VFW format, then you could use vdub, avidemux or any number of free open source programs to edit. You could even use WAX to do your PIP overlay. If you recall there are issues with WMV due to licensing and it's proprietary nature.

For encoding, there is a free h.264 encoder (x264). It must be installed separately, but there are many easy to use front ends, eg. ripbot264, handbrake, xvid4psp.

For decoding , there are several self contained players e.g. vlc, kmplayer, smplayer, which contain all the necessary codecs for hundreds of video formats. I think VLC even has a "portable" installation option, so there is no system install, it can come in a "usb" format.

A multipurpose "swiss army knife" codec suite is ffdshow, which contains many options for decoding and encoding, it's probably the most commonly installed codec.

If you want to keep it as simple as possible, you could bundle the apps or possibly write one yourself (all the software for my proposed method is free, most open source). WMV is certainly an option, but you need to instruct the end users to record at higher bitrates and fps, and most your quality issues will improve. Editing like PIP, overlays is difficult but can be done (especially with avisynth, but it's probably much to difficult for your users). Another factor for consideration is what kind of hardware do they have? WMV is very CPU resource hungry and you will be in danger of dropping frames (unless you use low quality, low fps or tiny screen dimension); this is because you are encoding (compressing) audio & video on the fly. But even old computers can capture with "proper" capture codecs without dropping frames.