Go Back   Steve's Digicams Forums > Digital SLR and Interchangeable Lens Cameras > Sigma dSLR

Closed Thread
 
Thread Tools Search this Thread
Old Aug 28, 2004, 3:25 PM   #11
Member
 
Join Date: Aug 2004
Posts: 34
Default

BTW, this is what made me wonder. The left sample is the Sigma...


Attached Images
 
code is offline  
Old Aug 28, 2004, 4:06 PM   #12
Administrator
 
Join Date: Jun 2003
Location: Savannah, GA (USA)
Posts: 22,378
Default

The reason that standardized tests are used, are so that users don't make the same mistakes as you are making.

For one thing, the conditions were no doubt different, which will impact image quality. In addition, the Sigma photos were taken with a Prime Lens (50mm f/2.8 ) and the Kodak photos were taken with a Zoom lens (24-85mm). We also don't know what aperture and focal length was used for the Kodak photos.

The most important difference, is that these photos were taken from two differing points of view, with a dramatic difference in the field of view (amount of the frame occupied by the building). The building occupies a much larger percentage of the frame in the photo taken by the Sigma.

If you look at it closely, you'll see that if youwere able to adjust the position you were taking thephoto from with the Kodak to match the position the camera was at for the Sigma photo; thencropped the Kodak photo so that thebuilding occupied the same percentage of the frame, you would most likely end up withless than half the original resolutionin the Kodak's image.

Remember, because resolution is computed by multiplying width x height, if you crop a photo to make it look like twice as much optical zoom is used, you end up with 1/4 the original resolution.

So that you can easily see the field of view, here are downsized versions from both cameras:

Sigma SD10:



Kodak DCS Pro/N:




JimC is offline  
Old Aug 28, 2004, 5:03 PM   #13
Member
 
Join Date: Aug 2004
Posts: 34
Default

I was aware of the differences in angle between the two, but not thatit was that big. The Sigma seems to cover approximately 6.5 Mp of the Kodak sensor's area.

And since it has 3.4 Mp at it's disposal and still manages to wipe the floor with the Daks 6.5, I still :love:this sensor.

And I don't care much about standardized tests either. I need to see the output for myself, which is why I bookmarked this site years ago.

code is offline  
Old Aug 28, 2004, 5:55 PM   #14
Administrator
 
Join Date: Jun 2003
Location: Savannah, GA (USA)
Posts: 22,378
Default

code wrote:
Quote:
I was aware of the differences in angle between the two, but not thatit was that big. The Sigma seems to cover approximately 6.5 Mp of the Kodak sensor's area.
Yes, there is a big difference between these images (and because they were taken from two different camera positions, it's difficult to say for sure how much the difference really is).

That's why it's a good idea to use controlled conditions when making comparisons. There are simply too many variables if you don't (lighting conditions, camera settings, angle of view, etc. -- not to mention the difference in lenses in these two examples).

code wrote:
Quote:
And since it has 3.4 Mp at it's disposal and still manages to wipe the floor with the Daks 6.5, I still:love: this sensor.

And I don't care much about standardized tests either. I need to see the output for myself, which is why I bookmarked this site years ago.
The conditions, angle of view, lens types, etc.were not the same (or even close).Your initialconclusions that the Sigma was capable of resolving more detail than a 14MP Bayer Type Sensor (when the building occupies a much larger percentage of the frame in the Sigma Image) only emphasizes why standardized tests are needed.

Look, I'll be the first to admit that resolution charts never tell the whole story. However, when making comparisons between models, you do need controlled conditions.

There is absolutely nothing wrong with loving this sensor. Many Sigma users do.

Lin already pointed out thatreviewers have foundresolution from Sigma SD9/10models isvirtually identical to models with6 Megapixels bayer processedsensor resolution. Lin was not trying to take anything away from thesemodels (and in fact pointed out that they are wonderful cameras).Lin only wanted to make sure forum users were aware of the facts:

Lin Evans wrote:
Quote:
Though I admire the enthusiasm of SD9/SD10 fanciers, the fact remains that these cameras have close to six megapixel bayer processed sensor resolution, but in no way do they remotely even approach 11 megapixel EOS-1DS Canon resolution which is considerably below the resolution of the 14 megapixel Kodak Pro line 14n/SLR/n range. They don't reach the eight megapixel resolution of the EOS-20D or EOS 1D Mark II either.

The SD9/SD10 are wonderful cameras which offer amazing resolution and some of the sharpest pixel level detail of any digital camera, but resolution is measured by an agreed upon standards calculated by lines pairs per mm and by lines per image height.Numerous professional reviewers have measured the resolution of the SD9/SD10 and found it approximately identical with the six megapixel Canon, Nikon and Kodak dSLR cameras and far below the highest resolution dSLR's.

As good as it is folks, let's not get carried away with exaggerations...

Best regards,

Lin
JimC is offline  
Old Aug 29, 2004, 5:52 PM   #15
Member
 
Join Date: Aug 2004
Posts: 34
Default

Thanks for the heads up, JimC. But I'm still not aware of the 'facts'. To me, a choice of digicam
is subjective. I decide what camera is best for my own use and preferences! Not industry experts.

What I do I zoom in and check if the image looks like it's real life. Like if you hear the news
report on your expensive stereo, and imagine a real human sitting right before you?

That's it, and no standardised tests in the world is going change the preferences of a camera once
it passes this one.

The closest thing I've seen so far to real life is, incidentally the Siggys sensor, and I expect
all sensors to be built this way in the future.

Also note that I sprinkled my initial post with some 'seems to' an 'If I didn't mess up's. I really
don't need to be brushed off as a fanatic based on that, do I. 8)
code is offline  
Old Aug 29, 2004, 6:18 PM   #16
Moderator
 
Join Date: Jun 2002
Posts: 1,139
Default

Quote:
Thanks for the heads up, JimC. But I'm still not aware of the 'facts'. To me, a choice of digicam is subjective. I decide what camera is best for my own use and preferences! Not industry experts.
Absolutely, the "choice" of one's tools or toys is absolutely subjective. On the other hand issues such as "capabilities and limitations" are indeed "objective" and subject to measurement.

"Resolution"is not a "subjective" issue but something which can and has been thoroughly investigated, measured and discussed on multiple forums.

No one should have any issues with your choice of cameras, lenses or equipment or your reasons for drawing your conclusions on what is "best" for you and what appears to give you optimal results. But, when you open the door to discussion about issues such as resolution comparisons which have been carefully investigated and results which have been widely published, you shouldn't be surprised when a statement which is far out of the bounds of reality is dissected and the flaws in the logic revealed.

We love the SD9 and SD10 cameras and the Foveon processor.It's a dynamite tool and has fantastic sharpness andproduces incredible resolution for the file size. But it doesn't even remotely approach the resolution of the 14 megapixel Kodak sensor you compared it with. You may "like" the appearance better, it may indeed be the ideal tool for your purpose and may produce thebest prints you've ever seen, but it simply doesn't do what you think it does and you should look at this as a learning experience rather than a"flogging about the head and shoulders" :-)

Best regards,

Lin
Lin Evans is offline  
Old Aug 30, 2004, 1:26 PM   #17
Junior Member
 
Join Date: Aug 2004
Posts: 2
Default


Objet : comparaison entre capteurs de type Bayer et capteur Foveon X3

Votre comparaison entre les appareils photographiques numériques 6 mégapixels et les SD9 / SD10 de Sigma manque de clarté et m'amène * réagir.

Je ne suis pas un théoricien des capteurs ni un professionnel de la photo, mais je suis très intéressé par le sujet et je pratique le numérique depuis dix ans sans pour autant avoir abandonné l'argentique, bien au contraire.

Mon propos se place sur un plan théorique et ne prend donc pas en compte les aspects de qualité de fabrication des capteurs ni les performances propres des appareils photographiques et des optiques utilisées.

Un 6 mégapixels avec capteur de type Bayer contient 1,5 millions de blocs d'informations formées de quatre capteurs RGGB ( 2 G pour compenser la faibleese du signal). Une image non interprétée contiendrait 1,5 millions de pixels, soit une définition de 1500 x 1000 pixels. Par un jeu de calculs, avec interpolations successives entre pixels et blocs voisins, l'image produites contient 6 millions de pixels avec une définition de 3000 x 2000 pixels, soit le quadruple de l'information recueillie. Cette interprétation se traduit par un certain effet de lissé et des artéfacts dus aux méthodes de calcul, mais le perfectionnement des algorythmes et l'augmentation de la résolution des capteurs permet de réduire ces anomalies.

Le Foveon 3 x 3,4 mégapixels collecte effectivement 3,4 millions d'informations globales RGB qui fournissent directement, sans interprétation, une image de définition 2568 x 1512 pixels : l'image obtenue est une représentation fidèle du sujet * la précision du capteur. Une interpolation analogue * celle utilisée avec les capteurs de type Bayer fournirait une image d'environ 5000 x 3000 pixels : cette possibilité n'existe pas, mais on peut toujours rêvé!

Au vu de ces considérations théoriques, ma conclusion est double :


-- comparer un 6 mégapixels de type Bayer et le Foveon 3 x 3,4 mégapixels est abusif ; si les définitions des images fournies sont voisinnes, par contre la fidélité de l'image, le "piqué" sont en faveur du Foveon ; un capteur 3 x 3,4 mégapixels Foveon se situerait en terme de qualité d'image entre les 6 et 15 mégapixels de type Bayer mais avec une définition moindre de l'image ;

-- les fabricants d'appareils photographiques numériques devraient fournir une information plus précise que la simple définition de l'image interprétée : une information "capteur de type Bayer avec 1,5 millions de blocs RGGB, image interpolée de 6 mégapixels" serait plus complète et moins trompeuse ; la publicité faite par Foveon et par Sigma répond mieux * cette exigence de clarté.
cigalaix is offline  
Old Aug 30, 2004, 2:04 PM   #18
Moderator
 
Join Date: Jun 2002
Posts: 1,139
Default

Quote:
-- comparer un 6 mégapixels de type Bayer et le Foveon 3 x 3,4 mégapixels est abusif ; si les définitions des images fournies sont voisinnes, par contre la fidélité de l'image, le "piqué" sont en faveur du Foveon ; un capteur 3 x 3,4 mégapixels Foveon se situerait en terme de qualité d'image entre les 6 et 15 mégapixels de type Bayer mais avec une définition moindre de l'image ;
No, it's "absurd" to compare the Foveon for image fidelity with such a wide range of bayer processed images. Six to fifteen megapixels is a difference of night and day. When real world images are compared from the Sigma SD9/SD10 with images from the Canon EOS-10D or Canon EOS-D60 there is great equity in "resolution" as well as in image fidelity and color fidelity by "measurement". When compared to even the "eleven" megapixel EOS-1DS, the vast differences in resolution are immediately apparent in favor of the eleven megapixel bayer processed image.

Perhaps when Foveon produces an array of six megapixel capture there will be an end to the arguments, but for now we should be very happy with the quality of our SD9 and SD10 images and not try to overstate the case for image quality or resolution beyond the reasonable which is precisely what has been done by numerous fanatical adherents of the technology. It's excellent technology and improvements are in process, but it has a ways to go before it will obtain dominance in the digital photography arena.

Best regards,

Lin

Lin Evans is offline  
Old Aug 30, 2004, 2:30 PM   #19
Administrator
 
Join Date: Jun 2003
Location: Savannah, GA (USA)
Posts: 22,378
Default

Well, I took French many years ago (but I don't remember ANY).:?

So, here's a Google Translation (http://translate.google.com/translate_t ) the previous post for the benefit of others (sometimes it's better than babelfish, sometimes it's not):

Quote:
Subject: comparison between sensors of the Bayer type and sensor Foveon X3 Your comparison between the cameras numerical 6 mégapixels and the SD9/SD10 of Sigma misses clearness and leads me to react. I am not a theorist of the sensors nor a professional of the photograph, but I am very interested by the subject and I have practised the numerical one for ten years without to have given up the silver one, quite to the contrary. My matter is placed on a theoretical level and thus does not take counts of them the aspects of quality of manufacture of the sensors nor the clean performances of the cameras and optics used. One 6 mégapixels with sensor of the Bayer type contains 1,5 million records made of four sensors RGGB (2 G to compensate for the faibleese signal). A not interpreted image would contain 1,5 million pixels, that is to say a definition of 1500 X 1000 pixels. By a set of calculations, with successive interpolations between close pixels and blocks, the image produced contains 6 million pixels with a definition of 3000 X 2000 pixels, that is to say the quadruple of information collected. This interpretation results in a certain effect of smoothed and artéfacts due to the methods of calculation, but the improvement of the algorythmes and the increase in the resolution of the sensors make it possible to reduce these anomalies. Foveon 3 X 3,4 mégapixels collection indeed 3,4 million total information RGB which provides directly, without interpretation, an image of definition 2568 X 1512 pixels: the image obtained is a representation faithful of the subject to the precision of the sensor. An interpolation similar to that used with the sensors of the Bayer type would provide an image of approximately 5000 X 3000 pixels: this possibility does not exist, but one can always dreamed! Within sight of these theoretical considerations, my conclusion is double: -- to compare one 6 mégapixels of the type Bayer and Foveon 3 X 3,4 mégapixels is abusive; if the definitions of the provided images are voisinnes, on the other hand the fidelity of the image, "piqué" is in favour of Foveon; a sensor 3 X 3,4 mégapixels Foveon would be located in term of quality of image between the 6 and 15 mégapixels of the Bayer type but with a less definition of the image; -- the manufacturers of cameras numerical should provide information more precise than the simple definition of the interpreted image: information "sensor of the Bayer type with 1,5 million blocks RGGB, interpolated image of 6 mégapixels" would be more complete and less misleading; the publicity made by Foveon and Sigma answers better this requirement for clearness.

I agree that there is a need for clearer standards (especially now that we have more than one sensor type being used in Digital Cameras). Reviewers have not only made resolution comparisonbetween Foveon Sensors and Bayer Sensor types, but have also posted comparisions of other types of images, too.

There is a need for some better and more standardized tests for things like color accuracy on a pixel by pixel level.

There is no argument that resolution chartsnever tell the whole story when comparing images from more than one model (I pointed that out in one of my previous posts in this thread).

Until we do have more agreed upon standards for representing all of the information these sensors are capable of capturing, we'll need to rely on comparing images taken in the same conditions from different cameras (and the above examples indicate the importance of having controlled conditions when comparing images).

Even if we have testing methods and standardsthat look at things likecolor accuracy, detail and dynamic range on a finer level, that still won't be enough. Users may prefer the way an image looks from one model versus another, regardless of color accuracy, etc.(so there will always be a human element that is subjective when comparingimages from cameras).


JimC is offline  
Old Sep 1, 2004, 12:46 PM   #20
Junior Member
 
Join Date: Aug 2004
Posts: 2
Default

Thank for your responses, Lin et JimC.

My english is not good : the Google's translationis better, but sometimes approximative.

Mon propos était provocateur, mais c'est bien l'objet d'un débat.

En pratique, je préfère actuellement mon EOS 500 associé * un coolscan de Nikon pour les bonnes images. Sur le terrain, je travailleen doubleavec un petit numérique (actuellement un Ixus 4 méga malgré sa forte distorsion évaluée * 5 %) pour un rapide résultat.

Je suis intéressé par l'acquisition d'un reflex numérique, mais je n'ai pas été convaincu par les performances observées. Je précise qu'une image se jugeavec sa sensibilité personnelle,ce qui est plus important que des mesures techniques théoriques. Le même genre de débat existeégalement avec les films argentqueset avecles optiques.

J'ai été déçu par les photo obtenus avec EOS 10D (image trop lissée), et j'ai donc refusé d'en faire l'achat bien que je sois équipé en objectifs EOS. Il parait que EOS 1D mark II est surpenant, mais son prix (4500 euros le boitier) etson encombrement(1,2 kg) sont trop lourds.

J'ai pu tester le Nikon D2H ; les résultats sont assez satisfaisants, sans plus,avec une définition d'image voisines du SD 10 mais le prix ( 3600 euros le boitier) et l'encombrement (1 kg) sont aussi trop lourds.

Par contre, j'ai été impressionné par quelques photo obtenus avec les SD9 - SD10 pour leur piqué (je ne connais pas le terme anglais correspondant) et leur réalisme. Ce sont celles qui m'ont donné les meilleures sensations. De plus, les boitiers Sigma ont une réputation de robustesse même si leur finitionparait sommaire .

A suivre ...
cigalaix is offline  
 
Closed Thread


Thread Tools Search this Thread
Search this Thread:

Advanced Search

Posting Rules
You may not post new threads
You may not post replies
You may not post attachments
You may not edit your posts

BB code is On
Smilies are On
[IMG] code is On
HTML code is Off
Trackbacks are Off
Pingbacks are Off
Refbacks are Off



All times are GMT -5. The time now is 11:00 AM.