GNU ZoranNT project: Zoran DirectShow capture for NT/2K

[Dr Mordrid]

Posts on the the Pinnacle forum point to work being done at sourceforge.net on a GNU licensed DirectShow Zoran ZR36060/ZR36050 NT/2000 driver.

http://zorannt.sourceforge.net/

So far they are at pre-alpha (0.0.0.2) on a European DC-10 driver only with plans for Buz support later.

This is necessitated by problems with the Pinnacle Win2K drivers akin to those with Matrox cards. Their Win2K drivers appear to be VfW and are having overlay & playback issues. Sound familiar?

Dr. Mordrid


[This message has been edited by Dr Mordrid (edited 27 June 2001).]

[Elie]

I wonder if drivers are created for Win2K/XP if they canbe ported to M-G400?
Hmmmm very interesting indeed.

Elie

[Dr Mordrid]

The sheer fact that such a project is necessary makes one wonder....

If Matrox's incompetence is the reason for not being able to come up with Marvel/RR-G Zoran/DirectShow drivers, then it must be pretty d**n contageous. Pinnacle also seems to have the same problem.

Judging from the derth of NEW Zoran MJPeg products based on DirectShow it's cropping up elsewhere.

The obvious question is: does this mean root the problem is not the Matrox/Pinnacle/??? programmers but the Zoran hardware itself?

If so then what good are all the temper tantrums and whining?

Just asking....

Dr. Mordrid




[This message has been edited by Dr Mordrid (edited 27 June 2001).]

[farid]

I'm trying to see the light here : how would that project fit into the Matrox paradigm where all Zoran chips are piggybacked onto the main video display cards ? It looks like this driver will only be for PCI-based cards.

Farid

[Dr Mordrid]

If one used their DirectShow filters a good programmer could, theoretically, write their own "Video Tools" for the capture portion of the card.

That assumes, of course, that they succeed where Pinnacle, Matrox etc. have failed.

Dr. Mordrid


[This message has been edited by Dr Mordrid (edited 27 June 2001).]

[wkulecz]

If the Zoran hardware is so crappy that it's impossible to write a working W2K driver for it Matrox/Pinnacle/IOMEGA still get the blame for buying into it and foisting it on the public. IOMEGA seems to get a bye, as they were the first to market and the first to drop it -- not even waiting for W2K to appear!

I've made this offer before (to ATI too over their dropping of AIW-Pro W2K support), Haig can have one of his programmers drop me the name of the DDK function that causes the showstopper problem and I'll look it up. If it rings true, I'll defend Matrox from now on and endeaver to explain the problem in plain english. Until then I'll continue to pan them for this. Looks like either way its a blunder -- bad hardware or bad software -- doesn't much matter if I'm robbed at knife or gun point.

Given the complexity of the W2K DDK and unless MS has changed their policy it costs $500/yr to get thru MSDN and last time I looked at the license (admidly a long time ago) its not redistributable, so I've little hope for an open source project succeeding. But if does, Matrox/Pinnacle is in for a boatload of ridicule!

Working Linux drivers would tend to prove there is nothing wrong with the Zoran chip.

I'd be thrilled with correctly working VfW codecs for the Marvel on W2K. That's what the Hauppauge card I'm using instead uses and its alot better than the nothing I have for my G200 Marvel right now. I'm hoping the final drivers will at least be usable with HuffYUV, but its the MJPEG back to tape that made me buy the damn thing it the first place!

--wally.

[Dr Mordrid]

"If the Zoran hardware is so crappy that it's impossible to write a working W2K driver for it Matrox/Pinnacle/IOMEGA still get the blame for buying into it and foisting it on the public."

Funny, the last I checked when those products were "foisted" on the public Win2K and DirectShow didn't even exist. They were brought out for Win95 then later updated for Win98/SE/ME, which wasn't too hard since all of these supported native Video-for-Windows. Win2K doesn't. It uses emulation for VfW support. Not *quite* as good and buggy.

That both companies have even tried to update their existing user base should be complimented, even in failure. At least they tried and didn't abandon their customers post-haste as Iomega did.

"Working Linux drivers would tend to prove there is nothing wrong with the Zoran chip."

There may well be nothing wrong with the Zoran chipset itself. My comment above really was pointed to the fact that no DirectShow SDK is available for the Zoran hardware than implying that the silicon is faulty.

The problem is that interfacing said hardware to MicroCraps Win2K/DirectShow environment is a PIA absent such an SDK. With Linux there is no DirectShow. No Win2K/DirectShow = a lot fewer problems. Score one for Linux.

Dr. Mordrid



[This message has been edited by Dr Mordrid (edited 28 June 2001).]

[wkulecz]

W2K was in beta when all this stuff was introduced. It was still called NT5 and there wasn't supposed to ever be anything beyond windows 98 making the non-discovery of alledged NT5/W2K crappyness look more like a conspiracy to rip off the consumer by mass-marketing dead-end products.

If this is the position these weasel company shills are gonna take, then I strongly recommend that everyone stop buying all computer hardware and software until after windows XP is released! Its pretty much what I've done other than bigger hard drives and more RAM. I'm certainly not buying DVD-R stuff until after XP comes out and its updated or proven to work corectly with it.

If consumers don't revolt and demand better, then they'll keep getting only junk! I'm PO'd enough to keep my $$$ in my pocket for now and help keep the industry bleeding.

--wally.

[mikiem]

Hi All

Admitting my electronics & coding skills are currently below par, & for what it's worth and all that, taking a glance or three at the Zoran chip docs it would seem this is really just a fairly dumb chip -- hardly in the latest DSP class if you know what I mean. You feed it an input stream, give it a few parameters, & it outputs the stream compressed -- they also recommend it for digital still camera applications.

Initially then, at first glance, it would seem to imply that the chip would be oblivious to OS, DX or anything related for that matter, & I'm sure as far as basic operations go, that's true. The hitch I imagine comes in with video capture being time sensitive, i.e.: the chip has to do its job very fast, & DX isn't.

Originally DX was an attempt to short circuit windows code to closer approximate DOS game performance so MS could better penetrate the market. It can't equal 100% more direct access to the hardware as you'll find in something like well written code in the Linux environ.

So, the current software & driver problems might be stated as not being able to support hardware mpeg at fast enough speeds reliably, perhaps adding: using the given hardware circuitry.

This doesn't exonerate Zoran, since maybe the chip design is slower in operation then needed. Also, if this is like the design process involved in other cards, the card manufacturer relies on the chip designer for hard and software references, i.e.: C Media Audio, S3 (past), the TNT2 chipset and so on -- using the C Media example, the manufacturer buys the same basic chipset/card & software as everyone else, then tailors/tweaks it for their product.

In this case perhaps Zoran can be faulted for not providing this level of support, or maybe the initial stuff they provided falls short?

At any rate, however much anyone might despise VFW, it's unlikely it will go anywhere soon. Even the Osprey line, heavily influenced by MS as it is, provides VFW support. There arguably are fewer alternatives, perhaps it's likely to remain so, as plans to move to a purely digital world falter in practice, at least in the US.

As I'm sure to raise some eyebrows, digital HDTV is considered by many in the industry as for most practical purposes dead for now. Moving consumers to digital cameras & camcorders is not happening at the rate many would like, mostly due to the value recieved for the higher prices -- they're simply overpriced, often by 100% or more. [camcorders are a good example, where hi-8 can be had for $300 US -- going digital means only a difference in circuitry, substituting a compression chip, whereas the tape transport can be made cheaper]

Personally, my opinion is that there are no ideal solutions available for analog video today at less then pro levels. I don't see that changing much until DVD writable drives become common, some analysts predict within 1.5 years. Then it's predicted we'll see real alternatives to the VCR, there will be a market for captureing analog video, perhaps the largest yet, and we'll see some big advances in both hard and software.

Off the soapbox now
mikie