Very poor rendering quality for HD files in VS X4

[Ken Berry]

They are still AVCHD files with those properties, though wrapped in the .avi extension. I find that I can nonetheless play them on my PlayStation 3 and HDTV from a USB stick or external hard drive. The difficulty, of course, is when you want to burn them to Blu-Ray when they will be re-encoded to Blu-Ray specs.

[Peter100]

I have tried disabling Smart rendering, but this forces entire movie to be rendered. As a result entire movie shows the shuttering effect now. Also it has much lower quality when compared to original clip. I have attached 3 pictures that show loss of quality of rendered by VS clip when compared to original. In those pictures the left side is a screen shot taken when playing original clip, the right site shows clip that was rendered by VS X4.

Here are the links to the pictures:
http://www.4shared.com/photo/5nUmwUZt/Compare_1.html
http://www.4shared.com/photo/FKHAkp3G/Compare_2.html
http://www.4shared.com/photo/BJJqMV0N/Compare_3.html

I suggest you download the pictures and view them at full resolution (i.e. 1920x1080) otherwise you might get scaling artifacts.

I have not mention this before, but I get the same lost of quality even for completely static objects filmed from camera installed on the tripod. (i.e. no movements at all) This essentially eliminates any issues related to “structured parts moving in different directions and at different speeds” and leaves rather plain decompress – recompress process.

The reason I have chosen clip with lot of movement for this tread, was to demonstrate the flickering / shuttering artifact. This artifact is much worse and more unpleasant than lost of quality itself. This flickering artifact is visible even on clips with very small amount of movement. (i.e. very slow camera pans or zooms). It is bad to the point that unless you have completely static scene, it will show otherwise.

Although I have no proof, or way of verifying this, but it seems that rendered by VS clips have slower frame rate. The other possibility is that the frame rate remains the same, but individual frames are repeated few times during rendering process. This would explain why the movements in rendered by VS X4 clips appear to be jerky.

If any one who has the tools allowing for frame level advancement and count during playback, could look at the rendered by VS X4 clip (http://www.4shared.com/video/QD0j0sKB/Rendered.html) and report back to the forum what are his/her findings that would be great.

Regards,
Peter

[erdna]

Ya, but he is talking about AVCHD 1920 x1080 60i files.

And indeed ffdshow doesn't handle AVCHD. The only thing that looks better for me is to render the processed file as an high bitrate mpeg2 (50Mb/s) file. And even then, some players do a better job than others. Splash-Lite looks best for me with this video.

[Peter100]

Just wondering if anyone had a chance to look at posted comparison pictures or verify if the frames are being repeated in rendered clip.

[lata]

Hi

By using Video Studio you can advance the video frame by frame, click the digital clock then the arrows,
This shows each frame which seems to progress normally. I cannot identify any duplicate frames except for the first two, which has been noticed in the past when using Panasonic 700 files.

Post edited


I have been struggling trying to replicate your poor video files.
The ones I create look good and didn’t really know what I was looking for.
I have rendered about 16 video files and have to be very critical to pick fault

I have just downloaded the three BMP samples, now I see the light.
However I am unable to create anything as bad as that.
Like you I have taken Snapshots to compare the frames.
The white line in your image looks more like a saw tooth (its bad) and its that line I concentrated on.

Interestingly and you are not going to believe this but when I use your Rendered file, I assume this is the bad one, and I take a snapshot the image looks good, very similar to the Original video Snapshot.
In fact I have just downloaded this again just to make sure I am looking at the correct file.

I have just compared my Snapshots taken from your rendered file and they look better than your good images, the ones on the left.
I will e-mail the image to you.

I don’t know what is happening at your end but I just cannot make a bad video.

[Kingston]

After rendering TEST.VSP, I played back the video in Windows Media Player 11. I could see interlaced lines in the first video clip, whereas on the second clip I couldn't see interlaced lines. The second clip looked like it had been deinterlaced by VS. Other than that, the first and second clips played back smoothly. The second clip actually looked cleaner in the rocks.

The codec Windows Media Player used to play back the video was Mainconcept (Broadcast) AVC/h.264 Video Decoder.

[Peter100]

Here are two additional project files (i.e. original clip + project file) that demonstrate poor rendering issue.

In the “Test 2” pay attention to the posts at the bottom of a screen and their movements. You will see significant difference in how they are appear in rendered portion and in original clip. Also last seconds of a clip (with trees and grass) shows rather jerky side movements in rendered portion.

In Test 3 pay attention to the rendering quality of the boys head and his back (left lower corner of a screen). You will see blocky low quality rendering there. Also the movement of the trees and grass on right site of a screen is jittery when compared to the original clip.

Here are the links to the files:
Project file for Test 2 http://www.4shared.com/file/AS0pmJiH/Test_2.html
Original clip for Test 2 http://www.4shared.com/file/vDezQ3QO/00024.html

Project file for Test 3 http://www.4shared.com/file/2MyBwtRJ/Test_3.html
Original clip for Test 3 http://www.4shared.com/file/a0uPuRoJ/00005.html

Here is the link to the rendered files that I get when running the above tests.
Output of Test 2 http://www.4shared.com/video/gP8RqypY/Rendered_2.html
Output of Test 3 http://www.4shared.com/video/AYQaRdYO/Rendered_3.html
Screen capture of playback of output of a Test 3 file. http://www.4shared.com/photo/EpDYl5Wa/Rendered_3.html

Regards,
Peter

[lata]

Hi Peter

Sorry but I am still having problems in identifying the poor quality.
I can view the files in a variety of players and have to be very critical to find any difference, yet you seem to indicate a big reduction. You say “You will see significant difference in how they are appear in rendered portion and in original clip.”
Well no I don’t?

Yes I can see some quality issues, strobing on the train bars, but the same amount on both files.

My monitor is 22” Flatron widescreen and give excellent quality.

Are you able to test the files on another computer?
Are you able to save the files to Memory Stick (usb) and play on your TV

Can you also tell me what effect you used on the second clips in the projects.
I have tried to find these but my program does not show any filters applied, even though the thumbnail does show a small icon. Strange I know?

[Peter100]

Hi Travor,

Strobing effect is the one that I am refering to. It is quite visable on Test 2 train bars. On orignial clip I can see more train bars ( i.e. larger numer of them ) and their movement is rather smooth. On rendered clip appears to be much fewer of them and they move in strobing fasion.

I have used brightnes filter set +2. ( i.e .the smalles change)

Yes, I have tried rendered files on my 120 Hz Sony TV and with the same result ( i.e. much more visable strobing on rendered portion)

Just want to make sure that when you compare the results you look at :
a) 0024.MTS ( which is the origninal clip)
and compare to
b) rendered portion from a output from the project

Is this correct ?

Did you have a chance to look at files from Test 3 ?

[lata]

Hi Peter

Now I am seeing the light.
I think sometimes we see what we expect to see.
Viewing the Test 2 the posts run from right to left, then the rendered section run in reverse.
A bit like a Cowboys and Indians -- wagon wheels revolving the wrong way.

Ok now I know what I am looking for I see it immediately

Not sure about Test 3, that doesn’t seem as bad, but then the action is slower?

I am trying an assortment of settings for the renders but all so far create the effect.

[lata]

Hi Peter

When I view 00024mts in the timeline, last frame I see double power poles.
Same when viewing frame 00-00-02-10 I see two sign posts when there should be one

When I render to the settings below I see one power pole, sign posts the video does look better.

Use Make Movie Templates Manager to create a template to MTS types

MPEG Transport-Stream Files
24 bits, 1920 x 1080, 29.97 fps
Frame-based
(HDMV-NTSC), 16:9
H.264 Video
Video data rate: Variable (Max. 16000 kbps)
Audio data rate: 384 kbps
Dolby Digital Audio, 48 KHz, 3/2(L,C,R,SL,SR)

Set the Quality Slider to 100%
Set Two Pass Encode

[Doug2006]

Hi, I downloaded the project file and clip of the car driving the road. I rendered it in vx4 two ways. One was optimizer and one was same as first. I have watched them on a Asus 23 inch led back lite monitor and they look fine to me. I set vlc to loop play and watch ten to twenty times and can't seem to see any difference between the first clip on the time line and the second. If I was under the gun to come up with something I could say that 'maybe' there was a flicker in one part of the rocks as they went by but I would be hard pressed to be sure of that. After watching that many times they look the same. Doug

[Ken Berry]

It's the flicker in the rocks that we are all talking about. And on a big screen HDTV it would be much more apparent.

[Natal]

Looking at the original images the most obvious area of poor rendering is the sky. If you look at the original the sky is a fine graduation of blue from top to bottom, but in the rendered footage it has been converted into broad bands of uniform color.

That is the kind of artifact that is very jarring to a viewer IMO.

There are other things of course, but the sky is the thing I immediately noticed.

[Peter100]

It looks like some of you can replicate both mentioned problems (i.e. flickering and lost of quality) quite easily, while others have difficulty even noticing it. Although I would like to find out what is causing such of variance of results, I also realize that it might be very lengthy and time consuming process. I am also not sure that even if we come to some conclusions, Corel will even attempt to fix them for X4, since they have already release X5.

Interestingly I have tried rendering same files with X5 (trial version) and it does not seem to have the flickering issue. I still see blocky color gradation and artifacts, significantly less details especially in shadow areas etc. Please note that I install X5 on the same PC so I am using exactly the same hardware, drivers etc but get different results. This also further proves that described by me the rendering issues are related and caused by X4 rendering engine/algorithm alone.

I can only hope that Corel will have a look at this thread and issue a fix for all of us X4 users.

Peter