D/A Converter Quality

[Dennis Hull]

Is there a noticeable difference in quality of D/A chipsets used in various level (consumer, prosumer, professional or by brand Sony or Canon) DV camcorders?? In another community Keith mentioned S-VHS output from camcorders probably not high quality (paraphrase) and I now see if I just view DV tape on TV using S-VHS output from camcorder I could be losing a lot of quality inherent in DV tape if D/A converter on camcorder is not very good. Keith pointed out one way around this was to use firewire output into NLE and output that to DVD disc and play to TV from that. But, if one camcorder has higher quality D/A output I could maybe avoid that "extra" step. Should mention, I am looking to buy in VX 2000, GL2 price range.

[Mike Rehmus]

Good question. I don't know the answer because I haven't measured the outputs from the different playback units I have. I wonder if a vector scope and waveform monitor could measure any differences or whether it would take a golden eye and a great broadcast monitor.

I can see no apparent difference between the analog output of my DSR-300, DSR-20, PD-150, VX-1000, or Canopus DVRexRT or Canopus DVRex ( although the Rex, not the RT is known to have a noisy S-Video signal due to the inadvertent hetrodyning of two internal clocks). But I haven't actually taken the same program material and output it via those units to a broadcast monitor. But I do switch back and forth between the analog outputs from the Canopus and the DSR-20. I see no difference on my broadcast monitor.

Twice a week I tape an acting class and then play back the video directly from the camera. I switch between the 150 and the 300. The quality is very good as far as I am concerned and the only differences in image quality I have so far attributed to the better lens, optical block and DSP of the DSR-300.

Somebody have cabin fever and want to do a scientific evaluation of this issue?

[Dennis Hull]

It made sense that DV camcorders would skimp on quality of S VHS conversion circuitry and give a low quality output. So, I'm thinking if I spent $650 on a "magic gadget" DAC 2 converter (DV to S-VHS and component and vice versa) from Datavideo I would get vastly superior conversion circuitry vs DV camcorder conversion circuitry; vastly superior S-VHS signal and vastly superior picture on TV. But, my hopes were dashed today when I read on Datavideo website "DV camcorders have S-VHS sockets capable of high quality output." Also, person I talked to at Datavideo, although not well versed in their products did not think I would improve on S VHS from camcorder socket. So, I think you can put the cabin fever project into hibernation unless someone knowledgable thinks otherwise.

[Mike Rehmus]

<<<-- Originally posted by Dennis Hull : It made sense that DV camcorders would skimp on quality of S VHS conversion circuitry and give a low quality output. So, I'm thinking if I spent $650 on a "magic gadget" DAC 2 converter (DV to S-VHS and component and vice versa) from Datavideo I would get vastly superior conversion circuitry vs DV camcorder conversion circuitry; vastly superior S-VHS signal and vastly superior picture on TV. But, my hopes were dashed today when I read on Datavideo website "DV camcorders have S-VHS sockets capable of high quality output." Also, person I talked to at Datavideo, although not well versed in their products did not think I would improve on S VHS from camcorder socket. So, I think you can put the cabin fever project into hibernation unless someone knowledgable thinks otherwise. -->>>

The chipsets in the stand-alone boxes, like the chipset in the Canopus products are identical to the chipsets in the cameras. The computer environment can cause some noise on the analog video (Canopus DVRex, not real-time) but the conversion process in the RT version (same chipset) is immaculate.

The $650 just paid to purchase the chipset from Panasonic or Sony, box it up with connectors and give it a bit of power. Oh, and give everyone in the distribution chain some profit.

[John Cline]

The S-Video output of the D/A converter in my PD150 looks noticeably better than the converter in my TRV-110 Digital8 camcorder. This is feeding both with DV via the Firewire inputs and viewing the S-Video outputs. The PD150 has less noise and greater detail than the Digital8. (Needless to say, there is absolutely no contest when comparing the cameras themselves.)

To my somewhat critical eye, the output from the Canopus ADVC-100 is in the same ballpark as the PD150.

For what its worth, for these comparisons, I am using a Panasonic BT-H1390Y broadcast monitor, which is capable of somewhere in the neighborhood of 800 lines of resolution.

John

(By the way, S-VHS and S-Video are distinctly different things and the two terms are NOT interchangeable. S-VHS is a 1/2" videotape recording format, S-Video is a method of transferring analog video using separate chroma and luma information carried on a pair of individually shielded wires usually contained within one cable. While it is true that all S-VHS machines have an S-Video connector, obviously, not all devices with an S-Video connector are S-VHS VCRs.)

[Andre De Clercq]

I often read about the "D/A" convertor in these posts. I am not a purist, but in the formal electronics definition a DV to Y/C (s-video) convertor is much more..Only a small block is the D/A (digital to analog) convertor. The main parts are the DV decompressor (HW Codec)resulting in (digital) YUV signals and the color encoder (NTSC or PAL). Both are digital blocks and only the final step following those two parts is a D/A concertor which finally produces the analog signals out of these digitally processed signals (composite or Y/C). The performance of a D/A convertor is strongly determined by it's "reconstruction filters". Consumer version are often simple lowpass filters which can compromise output bandwidth and/or can still contain digital noise components which can interfere (moire) with other (digital) circuits in TV sets

[Dennis Hull]

Haven't yet found reading that takes signal from lense of camcorder through to output on TV CRT so piecing it together from posts and other reading. dre, your last post came right after I read Canopus paper on YUV vs RGB video editing formats.

Here is what I understand now: camcorder CCDs process light from lense in RGB digital; RGB then converted to YUV (NTSC or PAL format) digital and digital compression occurs. This "YUV" and compressed (5 to 1 I think in DV) digital signal and maybe some other digital information is what is output on firewire from camcorder. Is that correct understanding on my part based on your post and others who have probably been making same points but were over my head before.

[Mike Rehmus]

Only the high-end shoulder cameras have a 'front-end' video output. It is supposed to deliver all of the resolution of which the camera is capable and not to have passed through any recorder electronics. Detachable viewfinders also look at the camera head video.

Your signal path is essentially correct. The DV compression is about 5:1 but the proponents of the compression scheme will tell you that the compression artifacts are more equivalent to a compressed analog signal of around 4:1 or so.

Basically what we get is component video, stored and transmitted digitally. That's why the color is so clear. Sync signals are regenerated whenever we need to view the video and are not stored as part of the digital signal.

[Andre De Clercq]

That's correct Dennis. Just this FYI: the YUV "NTSC/PAL" formats used for DV compression, relate only to the video and synchronizing formats (60Hz/50Hz ...525 lines/625 lines...). Only when it goes out in analog format the real Pal or NTSC color encoding (burst..subcarrier..quadrature modulation ...) and finally the D/A conversion is involved.

[Andre De Clercq]

Yes Mike, the real enthousiasts claim even ~3 :1 as compared to a standard MJPEG compression. It fully depends on the image content, because the "macroblock" reuse and a few other tricks involved in the DV compression concept are content sensitive

[Mike Rehmus]

Yup, try shooting a slowly revolving bar head-on. The block generation is incredible.

The reason I have a problem with the compression numbers as stated by the faithful is that the 4:1:1 compression is not accounted for in their fervor.

That 4:1:1 does cost some color detail that is apparent in wide shots of fine detail. Like the leaves on trees.

[Andre De Clercq]

Mike, the "better performing" DV compression vs standad MJPEG does not relate to the 4:1:1 rasters...both are supposed to have the same rasters in this comparisons. You are absolutely right when stressing the impact of color resolution on the perceived image quality. The leaves on trees is a good example. The whole story about the (accepted) reduced chroma resolution is based on the limited ability of the human eye to resolve color details and thus "we don't need that much color resolution". There is more however. The leaking color (the green of the leaves), due to reduced resolution also affects the luma part and there we see and need resolution! Take a monochrome picture of the (leaves) screen, and you will see what the eye as luma part sees: the color leakage around the leaves adds to the "black" surrounding the leaves and this means the the B&W image (luma) is being reduced in quality.

[Mike Rehmus]

I understand that. My point was that the 4:1:1 pixel dataset is another compression that impacts the video quality of DV as you expanded upon.