Welcome to Doom9's Forum, THE in-place to be for everyone interested in DVD conversion. Before you start posting please read the forum rules. By posting to this forum you agree to abide by the rules. |
|
|
Thread Tools | Search this Thread | Display Modes |
9th October 2013, 07:52 | #21 | Link | ||
Registered User
Join Date: May 2009
Location: Belgium
Posts: 1,744
|
Quote:
Quote:
Ok, and when I will have recorded this kind of footage, what script should I use to compare them ? |
||
9th October 2013, 08:38 | #22 | Link | |
Formerly davidh*****
Join Date: Jan 2004
Posts: 2,496
|
Quote:
David |
|
9th October 2013, 09:25 | #23 | Link |
Registered User
Join Date: May 2009
Location: Belgium
Posts: 1,744
|
Interesting remark.
It seems that understanding VHS's working is not so simple actually. Anyway, I recall my main question : how to compare CVBS and S-video captures (and also 2 CVBS captures, one with dvd as intermediary) with Avisynth, except with ColorYUV(analyze=true) whose results are not easy to understand as Y/C levels are different in all pictures ? The subquery concerned the Philips's TBC efficiency. |
9th October 2013, 21:37 | #24 | Link |
Registered User
Join Date: Nov 2006
Posts: 773
|
There is a MSU tool called Noise estimation metric, you could use that to compare i suppose, have a look here:
http://www.compression.ru/video/quality_measure/metric_plugins/ne_en.htm Last edited by Mounir; 10th October 2013 at 08:31. |
9th October 2013, 22:03 | #25 | Link |
HeartlessS Usurer
Join Date: Dec 2009
Location: Over the rainbow
Posts: 10,980
|
Perhaps built-in Compare() func could provide some measure of difference between clips (PSNR),
or then again, perhaps not. Just an idea. EDIT: Retasked comparison with a Merge(A.blur(1.0),B.blur(1.0)) of both comparison clips. EDIT: Ignore, just another Brainlesss Stainlesss idea
__________________
I sometimes post sober. StainlessS@MediaFire ::: AND/OR ::: StainlessS@SendSpace "Some infinities are bigger than other infinities", but how many of them are infinitely bigger ??? Last edited by StainlessS; 10th October 2013 at 17:37. |
10th October 2013, 08:34 | #26 | Link |
Registered User
Join Date: May 2009
Location: Belgium
Posts: 1,744
|
Thanks guys, I'm gonna have look at this.
Except ColorYUV(analyze=true), I was also told about Histogram ; http://avisynth.org.ru/docs/english/.../histogram.htm |
10th October 2013, 08:58 | #27 | Link | ||
Registered User
Join Date: Sep 2005
Location: Vancouver
Posts: 600
|
Why do you want a machine to tell you which is better? Capture a scene or test pattern with appropriate content and look with your eyes.
Quote:
Quote:
|
||
10th October 2013, 10:19 | #28 | Link | |
Registered User
Join Date: May 2009
Location: Belgium
Posts: 1,744
|
Quote:
Even with very good eyes, some things are hard to do without tools (hardware or software), for example calibration. I'm not sure to understand what you mean, I just made tests with extracts and still have to capture the entire VHS. |
|
10th October 2013, 11:50 | #29 | Link |
Registered User
Join Date: May 2009
Location: Belgium
Posts: 1,744
|
I tested Histogram(mode="levels") which shows very similar results between S-video and Composite captures.
I also tests ColorYUV(analyze=true), here are the results on a frame (S-video on left, Composite on right) ; and another ; And finally I tested Compare(clip1, clip2, "", "compare.log") on the exact same portion of video (15 frames), here is the log : clip1 (= clip_filtered) is Composite, clip2 (= clip_original) is S-video ; Code:
Comparing channel(s) YUV Mean Max Max Absolute Mean Pos. Neg. Frame Dev. Dev. Dev. Dev. PSNR (dB) ----------------------------------------------------- 0 2.0545 -1.3906 51 -74 37.6727 1 2.0336 -1.4016 35 -62 37.8019 2 2.0766 -1.3972 55 -61 37.5089 3 1.9896 -1.3571 43 -54 38.0099 4 2.0342 -1.3658 36 -46 37.6620 5 1.9665 -1.3317 40 -47 38.0741 6 2.0110 -1.3624 42 -45 37.8413 7 2.0702 -1.3564 42 -49 37.4747 8 2.0049 -1.3173 50 -63 37.8216 9 2.0380 -1.3558 32 -40 37.6822 10 2.0190 -1.4192 40 -44 37.8700 11 2.0492 -1.4073 36 -38 37.7398 12 2.0689 -1.3935 39 -48 37.5861 13 2.0683 -1.4123 53 -43 37.6493 14 2.0834 -1.3746 37 -38 37.5346 Total frames processed: 15 Minimum Average Maximum Mean Absolute Deviation: 1.9665 2.0379 2.0834 Mean Deviation: -1.4192 -1.3762 -1.3173 PSNR: 37.4747 37.7286 38.0741 Overall PSNR: 37.7253 Code:
Comparing channel(s) YUV Mean Max Max Absolute Mean Pos. Neg. Frame Dev. Dev. Dev. Dev. PSNR (dB) ----------------------------------------------------- 0 2.0545 +1.3906 74 -51 37.6727 1 2.0336 +1.4016 62 -35 37.8019 2 2.0766 +1.3972 61 -55 37.5089 3 1.9896 +1.3571 54 -43 38.0099 4 2.0342 +1.3658 46 -36 37.6620 5 1.9665 +1.3317 47 -40 38.0741 6 2.0110 +1.3624 45 -42 37.8413 7 2.0702 +1.3564 49 -42 37.4747 8 2.0049 +1.3173 63 -50 37.8216 9 2.0380 +1.3558 40 -32 37.6822 10 2.0190 +1.4192 44 -40 37.8700 11 2.0492 +1.4073 38 -36 37.7398 12 2.0689 +1.3935 48 -39 37.5861 13 2.0683 +1.4123 43 -53 37.6493 14 2.0834 +1.3746 38 -37 37.5346 Total frames processed: 15 Minimum Average Maximum Mean Absolute Deviation: 1.9665 2.0379 2.0834 Mean Deviation: +1.3173 +1.3762 +1.4192 PSNR: 37.4747 37.7286 38.0741 Overall PSNR: 37.7253 Last edited by Music Fan; 10th October 2013 at 12:29. |
11th October 2013, 06:44 | #30 | Link | |
Registered User
Join Date: Sep 2005
Location: Vancouver
Posts: 600
|
Quote:
As I see it there are two possible advantages of running your video through the DVD recorder:
For the first it's easier to compare using a DVD test pattern. For the second you could find a scene where the camera is still and there are vertical lines, if that ever happens on the tape. Both effects are pretty obvious to see when there is an improvement (with appropriate content), and none of the analyzing filters you ran would be particularly helpful to measure them. |
|
11th October 2013, 07:16 | #31 | Link |
Registered User
Join Date: May 2008
Posts: 1,840
|
If you are looking for an opinion the pictures on the left are superior in every way to my 20/20 eyes (is that perfect?). The right looks over-saturated, look at the back of the neck on the guy on the right and the left cheek of the lady on the right. The edges are cleaner on the left, look at right elbow of the guy on the right. Left picture also has more detail look at the fingers on the left hand of the woman on the right. PSNR also agrees but that doesn't matter much. Your eyes may differ.
__________________
PC: FX-8320 GTS250 HTPC: G1610 GTX650 PotPlayer/MPC-BE LAVFilters MadVR-Bicubic75AR/Lanczos4AR/Lanczos4AR LumaSharpen -Strength0.9-Pattern3-Clamp0.1-OffsetBias2.0 |
11th October 2013, 09:37 | #32 | Link |
Registered User
Join Date: May 2009
Location: Belgium
Posts: 1,744
|
Thanks for these answers !
Eagles have better eyes, but for a human I guess it's near perfection Now that you mention the saturation problem, I see it, but I have to admit I don't see very well differences on fingers and elbow. A little bit but I'm perhaps influenced by your analyze. Anyway, the saturation problem justify I choose S-video. And as I have to return the dvd recorder to its owner today, I don't have time anymore to make further tests. It's corrected by glasses but even with them, I remarked in some situations I didn't see details as well as some others. |
11th October 2013, 10:22 | #33 | Link | |
Registered User
Join Date: Mar 2006
Posts: 1,049
|
Quote:
Nope, two signal are recorded at the same time but on different frequencies (thus you can say that this is frequency multiplexing). Chrominance is recorded in native way (as chrominance itself is AM/PM modulated) on "low" frequency, luminance is converted to FM signal and later recorded on different frequency (or rather part of spectrum) such signal is combined and feed trough write amplifier to write head (usually this is one head - combined head write/read). This is same principle as radio - multiple radio station at the same time if each of them use different part of spectrum. --- I've check schematics for this Sony SLV-E830 and you can easily recognize that Y and C are separate signals up to point where they are combined as CVBS and feed to video output - oscilograms clearly show this principle (there are visible colorbar signal with two part - L signal is for example on pin 25 of the IC201, where C signal is on pin 48 of the IC201. So this VCR can be modified to provide Y/C signals. I don't know any of the of VCR with RGB output as this require CVBC decoder (NTSC/PAL/SECAM) - there is no added value to have such decoder in VCR as required decoder already exist in TV. |
|
11th October 2013, 14:29 | #34 | Link | |
Registered User
Join Date: Mar 2009
Location: Germany
Posts: 5,769
|
Quote:
However, if you detach the concepts of VHS from the S-Video ... then you might be right. S-Video is not always recommended. The reason is that the signal has no synch, and in studios, where the cables are quite long compared to home use, any difference in length between the two cables could cause an error.
__________________
Born in the USB (not USA) |
|
14th October 2013, 14:24 | #36 | Link | |
Registered User
Join Date: Mar 2006
Posts: 1,049
|
Quote:
Yes, for sure it require modifications, perhaps adding same video buffers but this should be not complicated - if basic principle is that C and Y are recorded as independent signal paths and later they need to be combined before output them as CVBS then outputting separately Y and C is possible and doable. Probably some schematics for VCR is required, also datasheet for R/W video processor can be nice but even with scope some optimal source of signal can be found and used. (im considering only to add Y/C output as this can be most important for us - to capture and process digitized video data). I can imagine even further step as whole digital processing by capture video head baseband and DSP such digitzed signal to restore particularly precious recordings but this is out of scope for this topic definitely. |
|
14th October 2013, 16:15 | #37 | Link | |
Registered User
Join Date: May 2009
Location: Belgium
Posts: 1,744
|
Quote:
I don't understand very well what you mean. What would be different compared to my current capture method ? |
|
|
|