| Author | Thread |
|
|
12/31/2003 01:19:53 PM · #26 |
Sure let's just ignore the fact that Sony makes the best CCD sensor is in the world, Sony tend to make the lowest noise sensors of any company. If Sony cannot get a sensor to be low noise, then it is pretty much impossible.
Also it is quite obvious that CMOS sensors will be more noisy than CCD sensors, I am no electronics expert, but even I can understand that a low-power sensor gives you a higher noise to signal ratio. That means that it is impossible to make a CMOS sensor as low noise as a CCD.
I don't understand this perverse fascination with increasing the megapixel count, there is no way that the lens on the front of the 828 can possibly handle 8 million pixels in resolving power. I have yet to see a cannon EF Mount lens that can resolve enough detail for even the 10D. That said, I wonder how anyone can imagine that the lens on the front of the 828 can resolve enough detail for 8 million pixels?
I am sure that the 828 has certain advantages that other cameras do not have, but when Sony said that they were aiming at the pro-market they were taking on all the companies that use far bigger sensors, and a more reasonable megapixel count. With that in mind it is not unreasonable for the 828 to be compared to the Canon 10D. This is essentially the same mistake that Olympus have made with the new E1, they said that it was professional, and it is not even compared to professional cameras. That means that most reviewers feel that Olympus have overstated the powers of the camera. The same is true of the 828. Also another natural tendency is to compare cameras of similar price points, and the 828 is in the wrong price point to be compared with 717. |
|
|
|
12/31/2003 01:22:46 PM · #27 |
| It is possible to make CCD very low noise by using JFET or GaAs technology but the price would be huge to pay ,maybe 20-30 times more! |
|
|
|
12/31/2003 01:29:15 PM · #28 |
| Nikon do use JFET in the D2H, but only to partial effect :( |
|
|
|
12/31/2003 01:55:47 PM · #29 |
Originally posted by sn4psh07:
Sure let's just ignore the fact that Sony makes the best CCD sensor is in the world, Sony tend to make the lowest noise sensors of any company. |
Make that Fuji with the S2 Pro SCCD, it beats all other dSLR when it comes to noise (the Nikon D100, D1h, D1x and the Pentax *ist D all use Sony CCD's).
|
|
|
|
12/31/2003 02:50:36 PM · #30 |
Originally posted by Azrifel:
Originally posted by sn4psh07:
Sure let's just ignore the fact that Sony makes the best CCD sensor is in the world, Sony tend to make the lowest noise sensors of any company. |
Make that Fuji with the S2 Pro SCCD, it beats all other dSLR when it comes to noise (the Nikon D100, D1h, D1x and the Pentax *ist D all use Sony CCD's). |
Moire? S2 pro is poor... And if you smooth the other cameras images to the same degree you get -- noise free images :) |
|
|
|
12/31/2003 02:58:55 PM · #31 |
Originally posted by sn4psh07:
I don't understand this perverse fascination with increasing the megapixel count, there is no way that the lens on the front of the 828 can possibly handle 8 million pixels in resolving power. |
I agree. I'm not sure if increasing the MP size on a CCD camera would increase the quality much, because there would be more noise. But that's just my guess. If thats the case, I'm curious to find out what MP size is the optimal size for a CCD camera. |
|
|
|
12/31/2003 03:39:53 PM · #32 |
Originally posted by eswik:
...I'm curious to find out what MP size is the optimal size for a CCD camera. |
It's entirely natural to want more pixels; if you can use all 8,000,000 of them to good effect, you'll see awesome results. Take a look at the (full size) Canon 1Ds cample images at DPReview; I guarantee you wil be awed. The 1Ds certainly uses its 11.1 Mpx to good effect. It is possible to design cameras that can take advantage of 20+ Mpx; digital backs for medium format cameras have been produced with up to 22 Mpx (last I heard) and the cameras can take full advantage of them.
Where the Mpx race becomes absurd is in packing 8 Mpx into a tiny 2/3 inch sensor format. The pixel pitch is only 2.7µm, i.e. there are 9 pixels in the same space as one pixel on the Canon 10D. No wonder the noise is high! Further, the actual sensing area is only a portion of this, so in the end, the dimensions of the area sensing the light are only a few times larger than the wavelength of the photons involved!
Even if the sensor performed flawlessly, the lens would need to be manufactured to incredibly tight tolerances in order to avoid visible image problems.
As expected, it looks as though both the the sensor and the lens are contributing to the observed issues with the 828 images. No doubt, the camera does produce very nice images, but are they 8Mpx worth of nice? No.
In the end, we can't conclude that there is really an optimal pixel count for digital camera systems. It wholly depends on the optical system design, with larger sensors and optics having an advantage (up to a point) in ability to produce superior images, but a strong disadvantage in price, e.g. $1000++ lenses.
Message edited by author 2003-12-31 15:43:58.
|
|
|
|
12/31/2003 05:37:04 PM · #33 |
| the amount of pixels dont make the image. i have taken shots with canon D30 that look much better than my f707 pictures, and can be printed larger. i was using a $1500 lens... how does the fuji s7000 compare with its sccd to the F828 in noise?? |
|
|
|
01/01/2004 01:28:35 AM · #34 |
| And don't forget issues pertaining to visible diffraction (airy disks visible in print). As you cram more pixels into the same sensor (size), your diffraction-limited aperture gets larger and larger. For an 8x10 on the f828, you'll probably want to avoid anything smaller than f/5.6. Considering many lenses are sharpest (who knows about the f828's lens?) around f/8, this is very bad news. |
|
|
|
01/01/2004 03:37:59 PM · #35 |
//www.landscape.com/reviews/cameras/sony828.shtml
Maybe should read this?
Michael Reichmann,s Review on the 828. Interesting indeed...
Go to The Luminance Landscape site.
Message edited by author 2004-01-01 15:56:20. |
|
|
|
01/01/2004 03:55:28 PM · #36 |
I can't find anything with that link. ??
T
|
|
|
|
01/01/2004 03:57:26 PM · #37 |
|
|
|
01/01/2004 04:00:31 PM · #38 |
Go to Michael Reichmann site.
This works.
Message edited by author 2004-01-01 16:03:19. |
|
|
|
01/01/2004 04:04:02 PM · #39 |
|
|
|
01/01/2004 04:06:46 PM · #40 |
| Thanks Dale. Some fun reading there. And a good photographer to. |
|
|
|
01/01/2004 05:06:38 PM · #41 |
|
|
|
01/01/2004 05:19:42 PM · #42 |
| Thanks for sharing, jm. I'm holding on to my 717 for a while yet. I'm anxious to see shots taken with the 828 here, particularly JJ's. |
|
|
|
01/01/2004 06:00:58 PM · #43 |
| The problem with shots entered in challenges here is that any image that is reduced to 640 pixels is hardly an appropriate way to evaluate the performance of a camera. You'd be much better off checking out the forums at DPReview where people will post full-frame images, or 100% crops (like in Mr. Reichmann's review). |
|
|
|
01/01/2004 06:04:57 PM · #44 |
I enjoying viewing images on-screen at 100% and scrolling around checking out details. This way of viewing an image is in no way a less valid way of viewing images as viewing a printed image as Michael (the reviewer of the article) would have you believe. If noise is present in the image than there is always the potential of it rearing its ugly head in a given situation. Some of these situations are: viewing at 100% on-screen, making enlargements from cropped-out selections, and making poster-sized enlargements such as 12x16 or 20x30. Having these options available is important to a lot of us just like it has always been important to many film users. I was a little amazed at how much Michael seemed to trivialize the amount of noise in his test images but I was also a little surprised by how much more noise was present in his images than from other sample images I have gathered. Maybe there are discrepencies in the amounts of sharpening that were applied, even though Michael's images said minus sharpening. So I'm not sure what the deal is. He sort of sweeps the noise issue under the rug by suggesting that the noise reduction programs will simply take care of it. While these programs can fix quite a bit of the noise problems it is not the same as not having the noise problem to begin with. When noise eats into details like in some of the window shots from the F828 the areas end up being ruined of their details and still end up looking rough. I'm using Noise Ninja on most of my shots from my F707 for improved results but I wuld be much happier not dealing with that added step in my workflow. Having said all this, I would still love to use the F828 for awhile to judge it for myself with my own workflow. Except for the very limited RAW abilities I am still very interested in all of the camera's other features.
T
|
|
|
|
01/01/2004 07:12:34 PM · #45 |
Tim, you pretty much wrote what I was thinking after I breezed through the Luminous Landscape review. It appeared to me that the images from the 828 were not quite as good resolution-wise as the 10D images, and the level of noise was very high (at ISO 64!!). Michael did allude to S/N ratio being the real defining factor, but then ignored this for the remainder of the review. The S/N ratio for the 828 seems very low indeed.
So what did I take away?
1.) The resolution of the 828, at 8.0 Mpx, does not quite match what is possible with a 6.3Mpx sensor and excellent lens. There appears to be no benefit in resolution from going to 8 Mpx. I'll wait for DPReview to weigh in before reaching a final conclusion on this one, however.
2.) The noise level is incredibly high, and once shadow detail is lost in noise, no software can recover it. I would never accept the level of noise in the 828.
3.) Overall, the 828 seems very well built and for the most part has well though out controls and functions.
In summation I think Sony would have been much better off making this a 6 Mpx camera, and using a slightly larger sensor. The noise then would be manageable.
Message edited by author 2004-01-01 19:13:52.
|
|
|
|
01/08/2004 07:10:36 PM · #46 |
The store where i work received it's first 828 yesterday. didn't had the time to look at it until now but i sppent almost all of my day going somewhere in the store and taking pictures (it's a VERY BIG store. something like 280,000 sq.ft.) and i came to some conclusions.
There are 2 usable ISO level on this camera: ISO64 and ISO100
Above ISO200 neatimage can't do anything
You really want to use a tripod
The Chromatic Aberation problem seems minor to me. (in fact i had it only one time, it was done on purpose in extreme shooting condition. It was a window frame in the sun.)
the camera is really easy to use.
The image quality is amazing.
|
|
|
|
01/09/2004 09:06:29 AM · #47 |
Originally posted by kirbic:
Originally posted by eswik:
...I'm curious to find out what MP size is the optimal size for a CCD camera. |
Where the Mpx race becomes absurd is in packing 8 Mpx into a tiny 2/3 inch sensor format. The pixel pitch is only 2.7µm, i.e. there are 9 pixels in the same space as one pixel on the Canon 10D. No wonder the noise is high! Further, the actual sensing area is only a portion of this, so in the end, the dimensions of the area sensing the light are only a few times larger than the wavelength of the photons involved! |
While this is correct, it is possible to more efficiently manufacture the sensors, so that you could get a reasonable 8Mp array into a sensor that size. As far as I understand it, the sensor areas are currently only covering about 30% of the actual sensor surface, due to metal layers and other layout factors. By embedding the metalisation you can effectively increase the sensor area by about 60%, or triple your pixel count/ size but retain the SNR. (or is it early and I'm messing up linear and geometric scaling - the point remains the same, even if the pixel count increase is actually 9x))
Not saying Sony is currently doing this, but it will be becoming increasingly common to see very high pixel counts (by current standards) on really small sensors, with much better noise rejection, as there is currently a whole lot of dead surface area just waiting to be turned into active sensor wells, with more intelligent fab techniques. |
|
|
|
01/09/2004 09:31:04 AM · #48 |
Originally posted by jjbeguin:
This Camera is such an exponentially huge improvement over anything I owned before that I don’t have much to complain about yet. Being colorblind, chromatic aberration is a lifestyle, not a defectâ€Â¦ |
I guess we'll have to have a different set of voting criteria for you know ...
Ed
|
|
Home -
Challenges -
Community -
League -
Photos -
Cameras -
Lenses -
Learn -
Help -
Terms of Use -
Privacy -
Top ^
DPChallenge, and website content and design, Copyright © 2001-2025 Challenging Technologies, LLC.
All digital photo copyrights belong to the photographers and may not be used without permission.
Current Server Time: 09/10/2025 11:41:55 AM EDT.
