Zardon

Read The Article Here
__________________________

Every time a new graphics card is released we hear great claims about how the new video enhancements will improve our viewing experience or that the hardware is as perfect as it can get in the High Definition HQV Benchmark. Unfortunately, for many reasons most end users have to pretty much accept these claims as it is not really possible for the average consumer to set up two identical systems, side by side and see what the differences are.

Well we have with the aid of some specialist equipment and the results are fascinating, not only this but we have HD screen grabs and detailed analysis to show you which card delivers the best IQ........

Zelig

Link for the bmp of the nvidia image2 test points to the wrong place.

This reminds me that I should get my monitors properly calibrated, I'm sure they're affecting my image quality far more than my video card.

__________________
--

geraldmango

links seem fine here although those bmps take some time to download.

Great article guys, great article.

LukeK

very interesting read, took me about 30 minutes to absorb it, even now im not sure I understand it all.

Nice to see an analysis like this, just wish I could capture stills with powerdvd, some nice setup you got there.

brutusmaximus

Last week I was actually thinking how helpful it would be to see something like this, yet no one has really done it properly before.

Very informative. top job.

__________________

[hobo]eclipse

no analysis or screen caps of ATI enhancements?

__________________

Veridian3

Originally the article was intented to be default against default but when we started looking at extra features it was clear that NV were selling the enhancements as a great new feature which would improve IQ therefore we tested them. When we went to ati and asked what settings they felt would give the best IQ they responded saying the default settings. So that gave us two clear areas to test:

1) Default vs Default

2) Best vs Best

It just so happens that ATI feel their default settings are best for most scenarios. (And based on our findings NV should really think the same about theirs!)

Judas

IMO, losing Aero when playing a dvd/hidef dvd is pretty damn minimalistic...

It's not very common that i see anyone sitting there watching something without it in full screen (usually goes full screen on playback anyways)

And if your settings up a home theater system to watch, i doubt anyone would run it in windowed mode lol.

While yes, it unfortuneate that aero has to be disabled, (which is all automatic and rarely do i ever get the time to see it happen as it's so quick in the transitions), i can't see that being a big factor for comparison.

As it's been clearly shown that ATI's HD acceleration is slightly better using one stream, it's another thing that isn't all to commonly used to run multiple streams which obviously ati loses at.


I'm not sure, orginally starting to read through the article, i was sitting on the fence as to which would be better overall... and i read seemed to lean more and more towards ATI being overall the better choice for the "professional" looking picture as ati was far more consistant and detailed even with a bit of a cyan look going for it. And then the conclusion confused me lol. Just seemed out of place but that's just my opinion.

__________________

humonous

Says who? I watch movies all the time in a window while doing something else ! thats the joys of a computer. If I watch something on my ATI card all the other windows screw up and it drives me crazy. Why should we deal with popups and a 300 quid card looking like one that costs 50 quid? Thats a ridiculous statement and im sure as hell glad you don't review for DH as you have your head so far up ATI's arse its not even funny.

a bit of a cyan look going for it ? wtf? clearly you haven't even read the bloody thing. Its clear ATI are better in the darker range with more detail but Nvidia have some cool tricks up their sleeve too. I have never met such an ATI fanboi in my life. Ive been reading through some of your posts on this site recently and no wonder so many people pick arguments with you. You love to read things into editorials that arent even there in the first place. It is clear this article is not only subjective but its presenting a wealth of information I haven't seen anywhere else. clearly you just dont understand it, im sure not many people will be shocked about that.

Veridian3

In terms of Aero, i completely disagree. Firstly its one of the major features of Vista... no component should ever have to disable it. Secondly i think you underestimate the number of people who have a movie (or TV series) playing in another window when on their PC surfing the net or chatting on MSN etc. ATI have been messing about with a fix for months and need to get their act together imo. (there is a registry entry in PowerDVD which is ready and waiting for ATI to complete their side of the work)

On dual stream decoding, I'm sure that loads of movie buffs watch the extra features on their discs, they shouldn't have to suffer huge leaps in CPU usage when doing so.

Now the IQ, thats really the whole point of the article. We gave you our opinion... what we like and dont like as well as what is technically correct... that gives you a starting point but mainly we want to allow everyone to make their own decision about what is best. Thats why we went to extreme lengths to get these captures and make them available for everyone. No-one is right or wrong in this area... its purely a personal preference. (Having said that, your comments re cyan are a bit odd)

Alex

A superb read and i spent quite a while looking through the images myself. The technical analysis was absolutely brilliant guys. Not only was there the opinion of the writer but the analysis of an expert in the field (our own Z).

As for Judas, seriously man. If you don't like the conclusion, then so what? Its what Veridian3 feels - this is clearly a subjective IQ article, and what he feels might not be what you feel (you were maybe confused cause Zardon wrote a lot of the technical stuff but didnt have any input on the conclusion section - right?). I was more interested in the technical analysis myself, I learned quite a bit from that. The fact there were two people writing from different angles was a great approach, you should do that more often.

I thank you for doing it, because quite honestly ive yet to see something like that on a tech site. It was like I was reading a semi guide yet being able to make my own mind up as well. Not force fed crap like I see on so many sites.

Some images I prefer ATI some Nvidia, however the fact that nvidia enhanced mode seems to misreading some scenes means there is still some work ahead, although it has potential. I think they just need to lower some of the saturation a little to get ahead of ATI, although perhaps with a less than stellar screen it might help...

floppychops

just joined up to say thanks ! ive been reading your articles this year and they are great. I wanted to make the effort to say thanks!

Fascinating to see the differences. I think ATI are still leading the way, although sometimes they fall a bit short depending on the scene.

ChaosMinionX

I absolutely love this review, another excellent job guys!

__________________

Blacklash

I logged in just to thank you for doing this work, and providing the HQ BMP examples from each card. I wish more sites focused on the multimedia dimension of graphics cards.

I'd love to see if you could notice any difference in quality in still image viewing or recognition, and other formats like wmv, mpeg, and quicktime etc.

Anywho, good show and thanks for taking the time to do it.

__________________

Judas

I said common... I wasn't saying straight out and out that no one watches movies in windowed mode. Obviously for you it's a big deal, i'm just making the point that for the most, it's likely to be no big deal.

Sorry for making imo, a reasonable point. While i'll agree it should have been fixed a long time ago.

? perhaps you should read it more thoroughly?

And direct from the article on the very first comparison page, which seems to continue to run through the entire comparison. Keeps things looking more neutral, but it would appear ati would have a better, more superior image if perhaps they didn't pump up the cyan as often. Course it's just personal preference at this time, as i'm all for a WARM more saturated picture then anything. But i'm a detail fanatic, at which point i'd rather go with the quality/detail then more saturated, far to dark looking picture that nvidia's default settings provide. But to be a little more picky, alot of people have different monitors, and it's a bit hard to really see the difference correctly without a monitor/tv to really compare both on, that is of absalutely amazing quality. This is where i've a bit of an advantage, because as of yet, i've yet to find a monitor that reproduces colors/shades that matches the Dell 3007WFP. So going from some of the 24/22/20/19/17 inch monitors i've got of various screen types/qualities, it's surpriseing how that nvidia/ati screenshots showup completely different vs each other.



ATI fanboi? We bit harsh arn'tcha? I'd consider myself a powervr fanboi, not ati in any manner, apparently you just have a dislike for me as rep says otherwise. That's all fine and dandy, but i wouldn't go around swinging the atifanboi bat around much more. The article is packed full of good, excellent and very resourceful information. Good stuff that i've come to thoroughly enjoy about DH's articles as well as reviews. But like anyone else, i've my own opinions, however my opinions can easily slide depending on the information presented.

Completely disagree, i've read through the article twice, perhaps i've misunderstood or misread some of it due to being interupted several times while reading it. (one reason for reading it a 2nd time).

Your likely quite right for the majority that may be viewing this article. Completely agree aero is a critical visual enhancing feature in vista (even performance enhancing from what i've seen oddly enough). It'll be nice when ati finally get around to spitting out the fix, So when the fix is made available, would that be the weight needed to turn the tables towards ati's favor in the article? I'm just curious how much weight aero really deserves, obviously for me it doesn't hold much as i'd rather have functionality and quality in what i'm doing, then seeing the outsides or edges of the program look "translucent". Having both is just a smidge of icing ontop of the cake. (imo)

I'm more then likely underestimating this as well, it's just simply not a common thing at all to require dual streaming, course i would sure love to have the nvidia results then the ati ones. Huge plus for nvidia no doubt, but it's fairly iffy on the weight value again imo.

Completely agree, i had no serious beef with the article or the conclusion, i just was a little on the shocked side after reading through it as the final score count "seemed" to be pointing in ati's favor.

As i pointed out earlier, the cyan comment is related to what was mentioned in the article, that ati seemed to add a bit more a cyan flavor to the image to give it a more neutral appearance (this was mentioned several times throughout the article if i indeed read it right).

__________________

Zardon

Your point was actually not very reasonable. We are here to review products and featuresets, not to gloss over an issue which should have been fixed some time ago. It would not be very thorough of us to dismiss a problem like this out of hand if it affects even 20% of end users. obviously you think differently to us, which is fine, but as none of us have any hard figures via a survey it will be hard to tell. Hard line is, its broken, fix it.

Actually they aren't just pumping up the cyan they are using it in key areas to give a more neutral image which is actually a very clever technique. If they just "pumped" up the cyan the image would have blue tints to it, which it doesn't. Your personal opinion on ATI having "more detail" is correct to a point. This detail is gained by lower black levels in the three quarter and total density areas which means that fine details will show through more. We are already working with Nvidia on this as they found our findings immensely helpful. Their techs are right now working on altering curves for future revisions.


Correct, very few people have ultra expensive monitors, however even those who do very rarely calibrate them correctly. The Dell screen you mention is actually just high end for the consumer market, for professional video or design application it is actually barely entry level. It all depends on which scale you are going by here Judas. I have worked for 20 years in the design industry and we had specialised monitors which cost 5k each and had calibration guns attached to the screen so you could balance the colour and gamma depending on changing ambient light in the room as well as diminishing colours from the aging proceess.

Ok, but you are aware that these tests we produced aren't in fact through ANY screen, they are raw captures from the output channels of the cards. These are then stored and analysed by myself on a variety of levels. No screen is even in the chain of technical analysis. If you don't understand this then I can explain further (at risk of boring everyone to death).

Lets not get into this in this thread, If you have any problems with comments from the public let a moderator know. I will not tolerate nonsense from certain people ruining the positive conversations. This is not directed solely at you, however I want you to let this lie. If I see any more personal insults ill deal with them myself or one of my moderator team will.

This is exactly what we want, everyone to make their own opinions from the results. We decided to approach this from two angles, One from Stuarts point of view as an experienced tech geek and a movie buff and from my side with just pure technical analysis of the image breakdown.


Explain what you are confused on and I will respond, however if you are confused as to the conclusion not tying in totally with my technical analysis then you need to remember than the article has two sections a: subjective analysis (stuart views on what he is seeing) and b: technical analysis - which is what ive recorded and detailed. This isn't a hardware review with hard line frame rate figures, we are trying to approach a HD Video IQ article from a rather unique angle here with the opinions of one person and the analysis of another. Obviously you can have differing views from Stuart but the technical analysis is sound and isnt really open for debate.


Again im a little confused. There is no tables to be turned here, as Stuart detailed in the conclusion he happens to prefer the overall image that Nvidia are presenting, he explains how they are lacking (and we did in the enhanced section quite clearly I felt), however one of his most poignant remarks was that a mixture of both images would be the best. Which it would be. Nvidia actually AGREE with this and are working on fixing their shortcomings as we speak. I find this extremely positive, and I equally find it extremely negative that an aero bug with HD playback hasn't been rectified when it was reported eons ago by Stuart himself. Who is to blame? I have no idea, nor do I care. We are here simply to reiterate any and all issues to our public, you guys.

Aero is not "icing on the cake", Aero is a featureset that Microsoft set out a long time and it is an integral part of Vista. Whether you like it or not is irrelevant. A manufacturer of graphics cards should ensure it works fully and that it is as well supported as possible. This then lets the end user choose, not be forced into losing it due to a bug.

If in fact your comment of "icing on the cake" is true to you, then why not give nvidia a compliment for getting it right. Surely a cake with icing is better than one without. We think so, and I assume that all educated enthusiast users will have the same viewpoint, if not then I suggest they get another hobby. I don't accept long standing bugs, and we certainly won't ignore them either. Driver Heaven has been responsible for reporting and helping to resolve many bugs, I feel this is a part of our duty as "Driver Heaven", and we will continue to do so when we get allowed the opportunity. If we can pressure or embarass a company into fixing things which really should not be broken in the first place this benefits the community.


Thats fine, if you are iffy on the weight of both a lack of aero and a lack of accelerated dual streaming support then I suggest you are merely a casual HD user and will obviously be quite content with even basic playback. This article was not intended for this audience and was indeed geared for the enthusiast user who wants to analyse and get the most from his hardware on a huge HD tv or on a brand new high end screen. I can make an educated assumption that even a user who hasn't the highest end hardware would appreciate wider feature support and to learn how their image is being displayed, otherwise we wouldnt have wasted a month on this as well as a lot of money.

There was no final score. No product was given an award and neither was told to be rubbish. Stuarts views are that right now in the overall scheme of things that the colour on Nvidia is better and the featureset is wider supported as well as having less bugs. My technical analysis clearly showed that certain IQ tests were in favour of ATI, however hopefully people read the whole piece rather than just the conclusion page as it is really not fully indicative of the whole article.

Judas

just a fyi,

i wasn't refering to the article as a review of any sort... I was voicing my own view on the situations. The technical data presented is beyond sound no doubt about it zardon, No one is wrong, but rather some of the questions were related to that of V3's own opinions (which aren't wrong either).

I know there wasn't any final score, or anything of the sort.

__________________

Tigre Marino

Forgot to mention that nVidia does not support High Definition Video Decoding on Microsoft Windows XP. The test is only run on Windows Vista... ATI supports HD video acceleration on my XP PC running a Radeon HD2400 Pro card, giving very low CPU usage on my Athlon 64 3200+ running on my HTPC.

LukeK

so we have to use an outdated old operating system to get bug free support? clearly Nvidia are moving onto bigger and better things and ATI are taking years to get proper HD support.

Yeah, lets have tests on XP. in fact lets go for Win 2000 or 98/ME in the next one.

humonous

Good for you, but DH is a tech site which uses relevant operating systems to test. XP has long been ditched here in all the reviews, no point in including it in this one as far as im concerned. XP is the past really. get with the times !

Veridian3

I would have to agree with the readers above, XP is no longer a product which really should receive new features. ATI, Nvidia and most other companies should be concentrating on ensuring the most recent OS has the best features and experience... its how we move forward.

Having said that, this page seems to indicate Nvidia have acceleration, including dual stream on recent cards, working: GeForce Release 174

ianken

I find this comparison highly subjective.

I see you employed the noise reduction features. Why? NR for HD-DVD and BRD is pointless, it merely replaces random noise with deterministic noise, and destroys detail. IN a film like "300" where film grain was added in post production you'll actually change intended image.

To determine of color is rendered accurately: color bars and vector plot. Using photoshop and film content is not going to do it. It reminds me of the guy who calibrated his TV using "The Matrix" and when he got rid of the "too much green" wondered why all his other stuff looked weird.

To determine if black levels are right: ramp pattern and waveform plot.

The difference in brightness between the images is due to differences in how the two cards handle nominal range of video content. You need to deal with it differently depending on the display device in use. Nominal black for video content is "16." Depending on how the display or decoder is handling it this can look "too dark" or "washed out." It is obvious to me that the AMD and NV gear is handling nominal range differently.

Anyway, I just don't trust image quality analysis that does not employ test patterns. Video is very standards based. There are very well defined methods of determining if a video system is accurate and I didn't see that employed here.

My $.02

Supersparky

I noticed that no consideration was made about calibration of the video card with the display (or perhaps I missed it). Comments about color rendering cannot be made unless the proper both device and display have been calibrated. Perhaps Nvidia's darker rendering is only a result of miscalibration. This is why both manufacturers have adjustments on the card. Nevertheless, no big deal in my opinion.

However, there was one thing that stood out like a giant wart in the "expert's" review on "noise". I had to almost laugh. Movies, unless it's a high-budget sci-fi flick from George Lucas, are FILMED, not many are made with high-def digital cameras yet. Film has grain. The highlighted portions in the review (blue water) where "noise" was highlighted was actually film grain and not noise. Since film grain is part of the original frame or image, I would give the render of such grain a better review than one that merely smears over it thinking it's "noise". Sure, one can argue that grain is physical "noise", but it's not real noise, specifically artificially generated artifacts in the electronic process, is what real noise is.

This was one of the complaints (from production companies) of movies in high-def and films made on cheap film stock. It was the fear people would complain about the image being "dirty" because the high-resolution of HDTV being able to render film grain. The older the movie, the more grain is shown, as film technology has improved. Stanard NTSC wouldn't be able to show grain and thus the movie would be rendered "cleaner" in a side-by-side comparison with the hi-def version.

Treating grain as noise in the rendering process has the potential of smearing out small details. It all depends on the algorithm used. Nevertheless, the best algorithm is patented and is what the industry uses to "digitally restore" old movies. This technology is not in your video card, nor its driver software. It's quite an amazing technology too, it actrually makes the result look better than the original. Explaining how it works would take too much here. The technique is so advanced and requires a lot of computing horsepower that it requires a cluster of servers to work on.

So-called "noise-reduction" in consumer video should be avoided. I turn that stuff off and my image is better than an image with the small details removed or smeared together. Even though film grain moves around like noise, it still shows small details amongst the grain that noise-reducing algorithms take out or smear over all for the sake of a "clean" picture.

9 times out of 10, that "noise" is probably film grain. Removing it, generally means removing the small details.

Zardon

Totally Incorrect with regards to the technical analysis. The frames are captured directly from the GFX card to a file on a HD. These files are then analysed via direct pixel breakdown. There is nothing in this equation which relates to the monitor being used as the figures are reading from the raw output of the files. If you don't comphrehend this then I can go into more detail.

Actually there is noise in every film, even pure digital CGI to disc. You will find noise even will come into the equation from a hardware level on the output. I am glad you are laughing, because I sure had a chuckle on your George Lucas commentary. I am sure we could fire up one of the last Star Wars movies and analyse that, however there will be noise in those as well. (and they also suck!). There will be noise on the transfers and equipment used. We aren't analysing recording techniques or various production values we are studying how products DEAL with identical situations for consumers in a real world environment.


The artifacting in that particular scene is partially caused by the enhanced nvidia algorithm (this is NOT grain - its heavy artifacting due to issues with a multiscan correction algorithm - grain with a modern recording technique never appears in 20-250 pixel random deformations), this issue has actually been verified by our sources and there are improvements being made to the process. The process for multisampling the colour swatches and linear bands will be improved.

The rest of your comments are absolutely irrelevant to the majority of our testing. reads more to me like a beginners guide to noise techniques.

Veridian3

It is very clear that either you haven't read the whole article or you just don't understand the overall point of it.

There were 2 aspects.

1) Compare the image quality which end users will see when they buy each card, install it and the drivers and then fire up a HD-DVD/Blu-Ray. This is also known as default quality.

ATI employ noise reduction by default, regardless of the disc used so that's what was tested.

2) Ask ATI/Nvidia what enhancements they recommend turning on/off and retest. ATI asked us not to change anything. To quote them "Default is fine". Nvidia suggest trying the Contrast/Colour enhancements as well as noise reduction, starting from a figure of 60 and experimenting. These are the settings that both companies asked us to use. Also known as "best IQ". (I already explained this earlier but clearly you didn't read that either).


So to clarify...
1) Default
2) Best

Then we discussed our opinion on the results both subjective and technical, summarised and then allowed YOU to download each image and form your own opinion on what is the best IQ, and what each company needs to do to fix their image. (I'm sure you agree neither is perfect).

Screen calibration, test patterns... neither are relevant to the technical analysis (read Z's post) and our screens are calibrated with professional equipment for the subjective analysis (which is never perfect but as close as possible and significantly higher than the average enthusiast). The images are true digital captures of the card output at the documented settings.

Alex

where on earth did these two idiots come from? clearly they didnt even read the testing methods. I dont see any better way to test images by grabbing directly from the card then analysing the raw files. its the purest analysis I can imagine possible.

Great job Z and V3. probably some losers from another site jealous of the work you guys put into this.

brutusmaximus

Well said. I think the problem here is that there are really very few people on the net capable of going into such image detail as Z can and as such its confusing as its such a groundbreaking idea.

I mean whoever thought of grabbing the output from the cards, direct to file, then breaking apart the raw files? there is no monitor in the mix, no calibration to cock things up, and the signal is as pure as possible from card to file.

I mean its great we have people coming over here to debate it, but at least make some sense, what im reading is clearly people interested in video but quoting information which isnt even relevant as they figure the whole thing was handled by someone looking at a bloody calibrated (or not) screen to get the figures.

take some time to read the thing right, guys, its embarassing

__________________

humonous

moral of the story. Don't take on Z with regards to image work. He will pwn everyone, guy is an expert in the field. totally. You outta have seen some of his work before when he had his own design sites. amazing stuff. respect needs shown. end of story.

floppychops

yeah I had to join up to post something yesterday and im amazed how knowledgable and intelligent Veridian3 and Zardon are. Some of their hardware reviews are incredible. (skulltrail and recent SLI reviews for example).

The methodology is sound, I think people just dont comprehend it, as everyone is used to people making purely personal opinions looking at images on a screen. This is the total opposite of this article tho, I learned a lot from it and i know my friends did on the other forums I visit. Its a briliant article and its good to read that Nvidia are using the article to help guide them a little in the future. The company seem very proactive in improving their products lately.

I think ill hang around here but the articles take me some time to read and absorb there is so much in them. I had to read this one three times to work a lot of it out, maybe more people should read them slower if they are getting lost which is clearly the case.

OmegaRED

Don't care that much about the issue since I don't have HD but damn that was some great analysis, it must have taken a long time to examine the different photos knowing full well there'd be a lot of heat if you made a mistake. We appreciate the effort!

__________________