I'm sorry, but you've gone off the deep end here. The sample you're looking at was captured with a SAA7134 chip, so not only are there zero lossy compression artifacts, H.264 certainly wasn't used. The SAA7134 dates back to 2001.
This sample does look worse than the previous ones, but it's because of the other issues you noted (smearing, huge jitter/warping, chroma ghosting). There also seems to be rainbowing along that bit that you enlarged.
I would like to see a capture of the piece of video that was posted earlier when he was using the Digital8 camera for playback, but this time played back using the SLV-T2000 directly into the USB3HDCAP with no AVT in between. His initial sample on VideoHelp showed that this StarTech device does a good job at correcting line jitter itself.
I'm certainly not averse to corrections about aspects of the operation of that capture device. I never used used it, but going here on what I've read elsewhere. Yes, I've seen the owner's posts in videohelp, so I definitely would never use it myself or recommend it to anyone else for VHS capture. Those posts have the same problems seen here. I worked on three caps in this thread, all with the same problems. Perhaps other users have some ideas about how to get a useable cap out of the owner's setup.
Since field order is arbitrary, you could say that the Philips capture has reversed fields compared to the other one. In what way does the Startech have worse interlacing artifacts?
First off all, thank you so much for your contributions
Related with the capture problems I have also analyzed my previous captures in DV /Avi directly from the D8 firewire output and the same flaws are visible which could eventually mean that the internal D8 A/D converter have the same type of capture/conversion problems. As I have some additional old capture devices I tried also with a Dazzle DVC100 /usb2 with the last driver and again the same problems arise.
In this moment having tried almost 5 different captures devices all presenting the same problems, one could say they are all inferior quality, but I think at least the Philips saa7134 and the internal A/D from the Sony D8 camcorder are of at least average quality, I begin to ask myself if the problems detected are really induced by the capture devices or are source ( tape ) camera recording related problems as member msgohan suggests.
As so my priorities in this moment are , first select the capture device and second optimize the capture process for the selected workflow and third restoration / enhancement .
In this context and related to the first priority what is important for me now is to identify if the capture card /device is able to extract the source potential without damaging it and if there are problems intrinsically related with the capture device itself . Everything ( flaws ) that are source related problems (tapes/camera ) and that I can’t modify as they have been recorded longtime ago or operator related (me ) that comes from my inexperience to set up and optimize levels , options , etc will be second priority for now . I will have time after deciding which capture device to use, to learn how to optimize my capture/workflow.
In this context I need your help to identify if any of my legacy capture devices ( Philips saa7134, Dazzler DVC100, Hauppage win/tv 878/9 ) or the new startech usb3 could provide satisfactory /good results in analogic video Pal captures (video8 and VHS ). I have some other capture/bridge devices but they don’t allow losseless capture like advc -55 or moviebox 710 usb.
If none of this cards /devices could provide satisfactory/good results I intend to buy a new card / device that could be either a legacy PCI card For the XP PC, a PCIe card for a i7 Pc running windows 8.1 or an external device for any of them. However the availability of new and second hand devices in European markets are somewhat limited but we can still see some limited offer from Hauppauge, Black magic , Osprey . All alternatives cards /devices suggested in this forum or the ATI are impossible to get from european amazon and ATI second hand are also almost inexistent. The Aja Kona is to much expensive and requires a beefed workstation . I can afford if necessary to spend in a new card/device up to 250-300€/ 300-350 usd.
Until now I’m not completely unsatisfied with the startech usb3 . I have desinterlaced some images captured from different devices and compared quality of the individual fields images in what’s detail is concerned and I think the startech gives the more detailed image (even if is a very tenuous difference ,oddly I got the best detail with D8 with tbc off ) , even if some results could be arguable I don’t think it´s worse than the others quite the opposite. The problem is I’m only comparing between the equipment I own and it could seem reasonable to me because I haven’t seen any better.
As Member msgohan was interested to see a sample of the first clip but played by the SLV-T2000 direct to the startech without the external TBC I’m attaching such a sample
One additionally point can you tell me the script to use to get YUV and RGB histogram like yours. I have AviSynth installed as I use it indirectly through videocleaner GUI. Probably I could use it in VirtualDub
Hmm, how do you expect a shaky horizontal pan across curved near-horizontal lines to look? This is an artifact of interlacing in general (later you used its name, "interlace twitter"). So it's just a source problem for these parts of the recording.
Sanlyn, my apologies. This new capture doesn't have this particular issue, and I now see what you mean. The correction is SwapFields() in Avisynth. The problem isn't simply that the old capture was BFF, that alone would be fine, it's that the lines were actually out of order.
Having said that, this new StarTech capture is TFF. I wonder why it isn't being consistent. Did you change to the other driver for one of these captures?
Yes , the first captures were made with micomsoft driver V. 1_1_0_140_1 and the last ones with micomsoft driver v. 1_1_0_141_5. Never tried the original startech driver.
My thoughts on the two StarTech captures... Note that I adjusted the images below liberally. The SLV-T2000 capture's chroma is in a totally different place than the DCR-TRV238E's. I decided to change only the vertical alignment, because changing it horizontally didn't seem to actually line it up.
You must be logged in to view this content; either login or register for the forum. The attached screen shots, before/after images, photos and graphics are created/posted for the benefit of site members. And you are invited to join our digital media community.
If you right-click on the images, view each in a new tab, and flip between them, it's easier to see the differences.
DCR-TRV238E
+ All horizontal details including text are clearer.
+ Head-switching skew is partially corrected.
- Somewhat noisier, more "processed" look -- especially edges.
- A handful of image columns on the left are masked with black.
- On right-most edge of frame, chroma is inverted(?). Possibly correctable. This was mentioned on VideoHelp in other threads.
SLV-T2000
- Everything exhibits a sort of horizontal smearing. Numbers (26) and words like the small red text are harder to read. Objects like the four dark squares on the wall leave faint horizontal trails to the right.
- Inverse chroma "ghosts" to the right of objects (i.e. red is followed by green, green is followed by reddish.)
- Head-switching skew is uncorrected. The rest of the image doesn't seem jittery, so I wonder why the card wasn't able to fix this area like it did in the VideoHelp capture.
Yes, msgohan, the horizontal field correction is a good idea. Vertical correction didn't work for me either, but never had time to try other ways. I'll give your idea workout later. I'm also wondering if the original camera's shutter affects the aliasing and broken lines.
Thanks for your input on these clips.
Quote:
Originally Posted by AJAX1
One additionally point can you tell me the script to use to get YUV and RGB histogram like yours. I have AviSynth installed as I use it indirectly through videocleaner GUI. Probably I could use it in VirtualDub
Why use VideoCleaner GUI to watch an Avisynth script? Run the script in VirtualDub and add VDub's filters to the output, in one step. You'll have to watch histograms to see what's going on when you make your own adjustments.
These notes and script are for the "Test.avi" clip in post #17 (http://www.digitalfaq.com/forum/vide...html#post38205). I used Avisynth for basic level fix and cleanup, then Virtualdub for tweaks -- all of it could probably stand more tweaking.
If you don't have some of these Avisynth plugins, we can find them them. For levels alone and no denoising, all you really need are ColorYUV (built-in), and instead of the SmoothLevels plugin you can just use Avisynth's built-in Levels as:
Code:
Levels(16,1.2,235,12,250)
VirtualDub filters used were ColorMill, gradation curves, and Hue/Saturation/Intensity by Donald Graft. The .vdf filters are attached as filters.zip. All three filters have numerous settings, difficult to describe one by one. So I have posted images of the filters as I had them set up.
I've also attached Test_vcf.zip, which is a VirtualDub .vcf settings file. If you open the .avi or the avs script in VirtualDub, click "File..." -> "Load processing settings" and navigate to the the location of the unzippeed .vcf file, then click "Open". The .vcf file will load the filters in their original sequence and with the settings I used. The .vcf is attached as a 'zip' file because ".vcf" isn't an allowed file extension in the forum. If you want to see what the .vcf looks like inside, it's plain text file that you can open in Notepad.
To use a .vcf to load filters and settings, the VirtualDub vdf filters must be in the VDub plugins directory. In case you don't have these three filters, they're in "filters.zip", zipped with a subfolder for each filter. The filters themselves are .vdf files in each subfolddr. Copy the .vdf's into the VDub plugins folder.
Here is how the three filters appear when loaded into VirtualDub:
ColorMill settings: left = MIDDLE POINT settings, right = SATURATION:
Gradation Curve settings. left= "RGB" luminance levels, right = "Red": The small anchor points near vthe top of each diagonal line keep brightness and Red at safe video levels. The other anchor points are used to maintain a straight line in the middle and lower portions.
Don't assume that the same filters or identical settings will work for every scene in analog video. Analog is about as variable as they come -- levels and colors can change with almost every scene. If you want, you can change the Avisynth script and the VDub filter settings to check the different effects. All of these filters are mainstream, but settings for any specific video will likely differ.
For the T2000/Startech vlip in post #24 (http://www.digitalfaq.com/forum/vide...html#post38236) I had to use different settings. I used Camcorder ColorDenoise, Colormill, VDub's "Levels" filter, and gradation curves.
Frame 274, resized for 4:3 display (deinterlaced -- it's frame 137 in the interlaced clip):
This capture is very murky in shadow areas, even with the higher black levels. Something I could try here is a contrast mask to maintain shadow details. But didn't have time for that. The capture looks too yellow. But most of the reds look rather grim -- in the capture, all of them turned toward violet, especially on the big front logo sign. I noticed those odd reds with my old SONY VCRs, which I no longer use. My remaining SONY is a rebuilt 1990 era SLV-585HF, the only SONY I really liked.
