4:4:4 10 bit or higher for under 200 bucks ? PAL

[user58]

hello all
i reconized the very old compression methods 4:2:2 ect.
then i was actually wondering why the recorder vhs image was that bad

while i gone trough magicx, hauppauge,easycap,vhs2dvd,honestech and other programs or hardware
i figured out my usb device only support 4:2:0

and all of those programs rather record a maximum if 8000 k/bits
sometimes not even that you left behind a compression and have no raw mode

then searching for a device what actually can do 4:4:4 8/10 or 12 bits i ended up with the
Intensity Pro 4K (200 bucks)
but hey its 200 bucks after all

so is there a cheaper way then 195 $/€ ?
searching for a card or usb device?

[Blue_MiSfit]

Look on the used market. You might be able to find an old Blackmagic or Aja card with a composite interface.

[Sharc]

Hauppauge USB Live2 + Huffyuv capture VHS to YUV 4:2:2 for example.....

[user58]

4:2:2 is already compressed
its better then 4:2:0
but i would prefer 4:4:4
then no pixel is skipped fused or has the same color
depending on the situation


the problem with the AJA cards is that the cheapest card you can buy that has 4:4:4 10 bit cost 350 $/€
often more they have cards going up to 9000 $


the blackmagic card is around 200 $/€

seeing the prices going up to 9000 $ that might be cheap

but really ? for a old VHS device ?
then 10 years and older harder still having such prices ? weird

no1 has a cheaper solution for 4:4:4 ?

[Sharc]

I don't think that you will notice a difference when capturing VHS tapes to 4:2:2 or 4:4:4. If the quality of the VHS captures is disappointing it is probably due to other reasons than color resolution.

[Cary Knoop]

Hold on a second, VHS is theoretically 4:2:2 but actually effectively much less than that, you can sample with 444 but that would not do any good.
And may even make things worse if you want to deinterlace!

[Sharc]

VHS is analog video.
After capturing lossless with Huffyuv encoder I am getting
Video
ID : 0
Format : HuffYUV
Format version : Version 2
Codec ID : HFYU
Duration : 9 min 31 s
Bit rate : 56.2 Mb/s
Width : 720 pixels
Height : 576 pixels
Display aspect ratio : 5:4
Frame rate : 25.000 FPS
Standard : PAL
Color space : YUV
Chroma subsampling : 4:2:2
Bit depth : 8 bits
Scan type : Interlaced

4:2:0 planar is YV12 for DVD and HD BluRay for example (all for consumer use), or when you capture and compress/convert VHS on the fly to mpeg-2.

[user58]

it says that 4:2:2 and other are samples
and it says that only on 4:4:4 doesnt skip any information
https://de.wikipedia.org/wiki/Datei:Yuvformats444sampling.svg

is there maybe a device that can read the VHS data from the tape via the head

instead of going for the output signal ?


for example it says it only notice that a pixel is there but not its chroma/color then if the color was a different color that pixel is skipped and get the same color from the pixel before

i meant to reconize this as compression routine from past ? i could be wrong

the diffrens is very huge :
https://youtu.be/FBmxH55j254?t=463

[FranceBB]

the diffrens is very huge :
https://youtu.be/FBmxH55j254?t=463


lol we're on Doom9, everybody knows what sampling is, you don't have to link a video to it, it's the international encoding forum, everybody is skilled, I wouldn't have suggested you this forum the other day otherwise. XD
Besides every video on YouTube is yv12 (4:2:0 planar 8bit) so the comparison that guy made is pointless. There are screenshots of different sampling and the difference is much more subtle than the one of the video.



it says that 4:2:2 and other are samples
and it says that only on 4:4:4 doesnt skip any information
https://de.wikipedia.org/wiki/Datei:Yuvformats444sampling.svg



As I told you on Skype the other day, I said that the best thing was lossless HuffYUV 4:4:4 but only 'cause I rarely deal with VHS and I have no idea about what their original sampling is. Anyway the general idea is that if it's native yv16 (4:2:2 planar 8bit) or yv12 (4:2:0 planar 8bit), there's no need to encode it as yv24 4:4:4.



Side note:
[you can trust everybody here as they know what they're talking about, that's why I wanted you to register and I'm glad you did. You should stick around, it's a nice community and I post here a lot, in Avisynth usage mostly. Heck, I just noticed that I check Doom 9 more than I check Skype xD
Of all the nicknames you could have picked, why user58 though? XD]

[user58]

Hold on a second, VHS is theoretically 4:2:2 but actually effectively much less than that, you can sample with 444 but that would not do any good.
And may even make things worse if you want to deinterlace!

you might help me out with this question a little more ?

the most information that is around in the internet says that 4:4:4 is sampling all pixels

i might missinterpret those 2 informations :
1:
https://upload.wikimedia.org/wikipedia/commons/5/56/Chroma_subsampling_ratios.svg
2:
https://www.youtube.com/watch?v=K4586b8iR2Y&feature=youtu.be&t=89

going below 4:4:4 (16) it calls out the old problem of compression
the want a smaller file size for example
then the internet useally tells a story what information is skipped
for example 4:2:2 (8)
only the half of pixels
according to this if the next pixel had the same color all is ok by this method
if a pixel was there that detail is lost

so it would be an "if" situation
4:2:0 half the pixels again (4)
4:1:1 different processing but (4)

a other thing that improved the image quality for VHS was a progressiv scan (more pixels but since the frames are combined sometimes the video shows bars)



(question about VHS image quality)
yes vhs image quality was bad and the image quality (personal videos from the past) are already of bad image quality

but exactly this question of having a bad image quality leads why the main raw file must be of highest image quality possible

after this i can set a compression the transfer speed can be 140/mb and 2000 gb of storage (thats the maximum my disc can do)

going for the signal and only taking 4:2:2 it would already compressed i might be wrong ?

but even wrose are the vhs tools 8000 kb/s maximum 4:2:0 or badder and again a compression (for example mp4)
and that beging progressed from a already compressed/skipped pixels

thats a triple lose if it would sum like that
and leaves the 50 frames question open
they either use interlaced videos or single frame scans (25p)

better would be a image of both (50 frames) and interlace them in high quality
(https://de.wikipedia.org/w/index.php?title=Datei:Interlaced_Animation.gif&filetimestamp=20061205214625&)

well looking?:
https://www.videoproc.com/media-converter/img/deinterlace-dvd.jpg


i could be totally wrong and im free to hear any kind of criticism

[Sharc]

As I wrote, VHS is an analog format. It has no 'pixels', no luma and no chroma quantization. Only when captured it gets sampled, quantized, converted into a digital format with pixels etc.
https://en.wikipedia.org/wiki/VHS#Video_recording

In digital terminology, the resolutions of VHS sources are APPROXIMATELY:
- NTSC 333x480 luma and 40x480 chroma
- PAL 335x576 luma and 40x576 chroma
So in either case the chroma resolution of VHS is about 1/8 of the luma resolution only. Especially the horizontal chroma resolution is poor.

Capturing and converting to digital:
4:4:4 would have equal luma and chroma resolution
4:2:2 has 1/2 chroma/luma resolution
4:2:0 has 1/4 chroma/luma resolution

So even 4:2:0 still has a better chroma/luma ratio than VHS (which has 1/8 chroma/luma only due to poor horizontal chroma resolution).
Capturing to 4:2:2 would however preserve the full vertical chroma resolution of VHS. Going higher makes no sense, as the horizontal chroma resolution of VHS is poor.
One cannot recover what is NOT on the source. One cannot unbake the cake.

… going for the signal and only taking 4:2:2 it would already compressed i might be wrong ?
Compressed by color subsampling yes, but you won't be loosing anything for the reasons given above.

…. but even worse are the vhs tools 8000 kb/s maximum 4:2:0 or badder and again a compression (for example mp4)
If it comes out visibly worse than the original it is definitely NOT due to 4:2:0 subsampling, I would bet. I would look for other reasons. Real-time ('on-the-fly') encoders/compressors are often not very good, or one uses wrong settings.

…. better would be a image of both (50 frames) and interlace them in high quality
Requires a good bob-deinterlacer like QTGMC for 50fps, and - if needed - re-interlace to 25 fps.

Post a sample of your captured VHS source and someone might help you.

[user58]

i appreciate your help (yes bobbing would be my first choise having the current information)

"motion detection" vs "others"
https://youtu.be/YczLRshnxQ8?t=236

"progressiv scan" vs "deinterlacing Interpolation"
https://upload.wikimedia.org/wikipedia/en/c/ce/Interlacingani3.gif

there is a lot information on the internet saying different things
or rather they both right depending on what question is made, some information seems to be missleading

for example there is info that the "pixelrate/sample rate/frequency modulated"is 3 mhz (or precise 3,375 MHz)
but then saying thats useally not the case and and it is common to use use 13,5 mhz for this (ITU-R BT 601)
on some other website it says DVCAM (DV SDL) from sony use 12 MHz for this

then there is the 50 hz or 50 frames question
getting a "pixel momentum" would lead to a higher amount of pixels/colors

the most internet sources also say 4:2:0 is MPG and JPEG compression

the most internet sources claim 4:4:4 to be the best to give 1 example would be this 1:
https://en.wikipedia.org/wiki/File:Colorcomp.jpg


sure the picture is progressed by all of this stuff somehow
that might lead to the open the question to read the data/waves that are being processed directly from the tape heads
and emulate them together as an image
this would make certain the data was read 1:1

having equivalent 335×576 pixels comes with a new question why do i record the image in 720 x 576?
the chroma going 40*576 (i assume 576 is vertical/y?)
then the pixels would be some like 4:0,5:4 "pixel chroma information?"


i dont know much about vhs i not even know if the 50 frames are full pictures
if i assume they are lines and not full pictures that raise new questions

some information say they are frames called "even field/odd field" are they lines are they full frames ? are the 50 or 60 lines or fullframes ?

the pictures i took dont look different from the example i repostet
progressiv made the best quality (interlacing looked very smooth but lower resolution and lagging)
progressiv had the line artefacts

it looks like the colorcomb.jpg example from above

[FranceBB]

i dont know much about vhs i not even know if the 50 frames are full pictures
if i assume they are lines and not full pictures that raise new questions

some information say they are frames called "even field/odd field" are they lines are they full frames ? are the 50 or 60 lines or fullframes ?

the pictures i took dont look different from the example i repostet
progressiv made the best quality (interlacing looked very smooth but lower resolution and lagging)
progressiv had the line artefacts

it looks like the colorcomb.jpg example from above

They're probably 25i with rasters so you get even and odd fields; by bob-deinterlacing them with, let's say, QTGMC, you get 50fps progressive and that's the way it should be. Leaving aside the whole phosphors story of CRT TVs, modern TVs are all progressive and whenever they get an interlaced signal, they bob-deinterlace it on the fly, so you should bob it.
Interlace was made because CRT TVs displaying rasters were based on a beam hitting the screen and making raters brighten up (you can think about rasters as ancient pixels). The problem was that the beam could hit only a few rasters at a time, so doing it progressively would have led to flickering as when the last raster (bottom right) was lit, the first raster (top left) was already fading, so they split the screen in 50 lines, 25 even and 25 odd that were refreshed every time, so you could think at the Even lines displaying the current frame and the odd lines displaying the immediately successive frame. In Europe it was 50 times per second 'cause it was easier to make it work at the same rate as the monophasic current reaching user homes and in the U.S it was 60 times per second for the very same reason. Nowadays, interlaced doesn't make sense as TVs have been able to display pixels for years and, you know, all the pixels bright up at the very same time all together, so there's no need for interlacing and every display is progressive, therefore you should bob-deinterlace it to 50p using both Even and Odd fields to make a progressive frame, otherwise your TV will do it for you.

[Sharc]

the most internet sources claim 4:4:4 to be the best to give 1 example would be this 1:
https://en.wikipedia.org/wiki/File:Colorcomp.jpg

Sure, but again: 4:4:4 is total overkill for VHS. You won't get any benefit from it. Remember that HD blu-ray uses 4:2:0 only.

that might lead to the open the question to read the data/waves that are being processed directly from the tape heads and emulate them together as an image
this would make certain the data was read 1:1
The problem is the tape itself (magnetic particles) which limit the true resolution, not the heads. You cannot recover from the tapes what is NOT on the tape.

having equivalent 335—576 pixels comes with a new question why do i record the image in 720 x 576?

Mainly because 720x576 is a standard (typically used for PAL DVD) which has its roots in digitizing analog TV. It's actually already overkill for VHS which has a significantly lower horizontal resolution.

i dont know much about vhs
I agree.

[Asmodian]

I have done a lot of VHS capture going for maximum quality.

4:2:2 is very important for capturing VHS. VHS is analog lines of video so you do not want to mix lines. There is real separate color data samples for each line but not much per line. 4:4:4 is completely pointless, even 4:2:2 is a lot of extra samples of the color data.

4:2:0 blurs these lines together (sub-sampled vertically) and because we capture as fields (VHS is interlaced) this ends up mixing lines that are actually an entire line apart. For this reason it is much better to capture in 4:2:2 and covert to 4:2:0 after deinterlacing.

[Sharc]

I have done a lot of VHS capture going for maximum quality.

4:2:2 is very important for capturing VHS. VHS is analog lines of video so you do not want to mix lines. There is real separate color data samples for each line but not much per line. 4:4:4 is completely pointless, even 4:2:2 is a lot of extra samples of the color data.

4:2:0 blurs these lines together (sub-sampled vertically) and because we capture as fields (VHS is interlaced) this ends up mixing lines that are actually an entire line apart. For this reason it is much better to capture in 4:2:2 and covert to 4:2:0 after deinterlacing.
Yes, VHS shoud always be captured in 4:2:2 because of the interlaced format of VHS.
If one re-encodes the interlaced captured footage and wants to keep it interlaced (for whichever reason) it should still be re-encoded in 4:2:2. x264 for example supports interlaced encoding with the profile High 4:2:2. Unfortunately many HW players (TVs) seem to reject this format, so one has to encode interlaced (MBAFF) in 4:2:0, or deinterlace and then encode in 4:2:0 as you recommend.

[user58]

for those who are interested

progressiv scan gave a far better resolution
but have this "rills"

the record left is progressiv scan
the right is Interlaced Frames

[Emulgator]

Your screenshot shows 2 different players. MPC? MPC-HC? vs WMP ?

You can neither trust nor compare these rendered results anyway.
A lot may happen differently in between decoding and display output.
Look for the red print "SWR" on the T-shirt: not even the same color.

I would suggest to use VirtualDub for such comparison.

(The "rills" as you call them got to be there after capturing.
If these rills are smeared in the captured file something went wrong.
We take care of those rills after capturing.)

[Cary Knoop]

for those who are interested

progressiv scan gave a far better resolution
but have this "rills"

the record left is progressiv scan
the right is Interlaced Frames
For the best quality, you must capture the video as interlaced.
You can always deinterlace later if you want to do that.