Sony A350: Full-Time Live View at 14.2MP

113 Comments

View All Comments

• danddon - Thursday, April 3, 2008 - link

  I have to say that this forum thread is much more interesting and entertaining than the camera test has been.
  
  First, thanks for fixing the a350 pictures. They now look more like the other two sets in terms of lighting. The a350 results in the new pictures look more like a DSLR than a $50 Vivitar.
  
  Now, about the a350 versus 5D comparison: Does anyone else think that comparing an APS-C camera to a Full Frame camera a complete waste of time?
  
  Look at the crops. Since this is a visual comparison, the different sizes of lettering and the relatively larger amounts of black space in the 5D crops make any sort of comparison difficult at best.
  
  How about – as a minimum – use a 75 mm lens on the 5D, or at the very least, a zoom lens set at 75 mm. That way the details in the crop areas would look much closer in size.
  
  Or, failing any of the above, select a photo target with enough white space (and no lettering) to negate the differences in image dimensions. In other words, both crops would be all white. I didn’t see any white boxes in the photos shown. Perhaps the budget doesn’t allow for the purchase of such high-tech testing gear.
  
  Or, how about a GretagMacbeth color rendition chart, so that noise can be seen for different colors, and not just black and off-white, plus a little green? BTW, the black area in the 5D crop looks completely noise-free, but, is it? In other words, how does one look at those crops with all the black and see how much noise is there?
  
  In the meantime, we all now know a little bit about how the a350 fares when taking 100 watt light bulb pictures of a collection of boxes in Wesley Fink’s office, but not much else about the camera. I suppose we should be thankful for that much.
  

• Wesley Fink - Thursday, April 3, 2008 - link

  The usual ISO noise argument goes that the full frame has a tremendous advantage at the same resolution as an APS-C because of the increased size of the individual pixels in the full-frame image. I only included the 5D to try to illustrate the same pixel crops, as all are 230x300 pixels. Yes, I could have used a 75mm lens from the same position or I could have moved the tripod closer to the packages for the 5D.
  That brings up the next question – do you make the scene always the same “image area”? If you do you adjust the tripod distance with almost every camera tested, or you use a slower zoom lens that is likely not in its sharpest aperture range at f4. Even two cameras with the same claimed 1.5x multiplier will in reality show different views as the 1.5x is an approximation. With a constant position the images show the impact of the actual multiplier.
  I could have used the White “ESA” lettering next to the selected crop area or the large white label on the next box or the large white P5K Deluxe lettering. There are plenty of white choices for crops in the larger image.
  

• danddon - Thursday, April 3, 2008 - link

  Well, excuuuuuse me.
  
  I thought this review was about a visual inspection of noise levels. I guess I was wrong.
  
  The a350/K20D crops like fine for the purpose, because the focal lengths are the same, and the sensor is the same size. The crop is of an area that contains a white background with black lettering. Noise can be seen not only in the background, but also at the boundary between the background and the lettering. However, can you see noise within the black lettering itself? I can't - at least not without additional digital manipulation of the crops.
  
  Now consider the a350/5d crops. Are you saying these are equally effective at comparing noise levels? Because - if you are - then you should start looking for another job. Perhaps you could go back to being a "Pro" photographer.
  
  Apparently you _do_ think they are as effective, otherwise you would have chosen a better crop location, or adjusted the lens focal length, or moved the tripod, or, or ...
  
  I just hope someone in management at AnandTech is reading these posts.
  
  And, to answer your question - no I don't always "make the scene the same image area". Only when I am trying to compare noise levels from two cameras. All of the other review sites seem to make the same mistake, too, especially "The Imaging Resource", which goes to great pains to make all of their photos as similar as possible.
  
  Its too bad they didn't check with you first to learn the correct technique.
  
  
  
  

• jake123 - Thursday, April 3, 2008 - link

  Honestly I feel that some of the commenters simply want to prove others wrong to boost their little self-confidence.
  
  We don't need another dpreview. I find the perspective of this article refreshing, without irrelevant detail and I think this is what the readers need.
  
  Sure the review process can be improved but anandtech does not need to become another dpreview.

• Hulk - Thursday, April 3, 2008 - link

  "Irrelevant detail" in this case = accurate testing methodology
  
  

• ElFenix - Thursday, April 3, 2008 - link

  accurate testing such as having the D300's noise reduction on while having the 40D's off?

• Hulk - Thursday, April 3, 2008 - link

  Wesley,
  
  First I want to thank you for your hard work with these camera reviews. You (and Anandtech) are moving into somewhat new territory with dSLR reviews. As you are learning there are many people using dSLRs that are very knowledgable, even if they are pros. My Mom is a professional photographer with brilliant composition skills (in my and many other's opinion) but she couldn't do a proper camera review/comparison if her life depended on it. Just because you can take a good picture doesn't mean you can do a good review. Please realize that many people have little patience with the learning curve necessary to get up to speed with many established, cough,,, cough, dpreview, cough, photo review websites. Some of the people here need to take it easy on you and refrain from personal attacks and stay to the facts.
  
  In the future I would recommend resolution and noise testing be done in full manual mode so that we can see how each camera imager exposes at various settings. Also please keep f-stops constant and at least at F/4, higher would be better to remove lens quality from the testing variables.
  
  Also please use manual white balance as it adds another variable to the resolution and noise testing. Automatic white balance should be a separate part of the review and need not involve resolution or aperature actually. In fact, most people buying these cameras, when doing mission critical work use RAW format and white balance in the image editor. Also when doing automatic white balance testing please make sure you test each camera under a variety of lighting temperatures. As many people will agree automatic white balance is somewhat like "watching a dog walking on hind legs, it's never done very well, but it's amazing it's done at all." That is a quote from "Copying Beethoven." Have a look around the web at some technical articles on it and you will see the inherent problems with it.
  
  Finally, please do all resoution/noise testing with RAW images and use the same settings for conversion to TIFF files.
  
  I would suggest thinking of these camera reviews like you guys think of overclocking, philosophically of course. That is "isolate and consolidate. The biggest problem is too many variables in the testing. Define what you are testing and then hold as many of the variables constant as possible.
  
  Again thank you for the hard work and I look forward to seeing the Anandtech camera reviews becoming among the best on the web.
  
  - Mark

• whatthehey - Thursday, April 3, 2008 - link

  You know, all the stuff you mention is exactly what I DON'T want to deal with in a camera review. I want to know how well the camera works with as little effort as possible. Manual white balance in particular is way more effort than I'm willing to expend. Granted, I'm very much an amateur photographer (if that), but I do appreciate the ability to change lenses plus the almost universally faster AF speeds on a DSLR.
  
  Considering this is more of an entry-level model rather than a $1500+ pro model, this sort of review covers a lot of what I want to know. Could it cover more? Sure. The difficulty is in coming up with a good battery of tests where the photographer's skill won't skew the results. For example, I like a camera that I can set to ISO 100 (for low noise) and still get a good quality picture without always resorting to using the flash or a tripod. I'm also far more concerned with auto WB working well than with what can be done using the manual WB.
  
  DPReview does cover a lot of good material, but they are almost too much. I mean, 29 pages on the Canon 40D as an example. I can't read through all of that! Maybe if I were in the market for a new DSLR I'd feel differently, though.
  
  Really, I'm interested in a short, quick overview of a camera. Is it better or worse than the competition? Does it offer anything revolutionary that would make me want to switch? I get a reasonable feel for that from this review. Reading the DPReview article on the same cameras, I get... NOTHING! Probably because they're still working on putting together several new 30 page magnum opus articles looking at the A350, K20D, and 5D used in this review.

• Hulk - Thursday, April 3, 2008 - link

  Well I can understand you not wanting a review that is too in depth. That's fine. Although I find it a little strange when you are talking about a 14Mp dSLR with removeable lenses.
  
  What I can't understand is not wanting to have testing done in a scientific manner so you know what is actually being tested. For example if you are testing resolution and one camera is set to f/1.4 and another at f/1.8 then that will skew the results significantly.
  
  But as you say there are probably lots of people like you that don't know or care to know about aperture, latitude, or a variety of other photography related terms.
  
  BTW, the pages on dpreview are labeled so you can go to the page you want. You can always just skip to the conclusion to read a well throught out and supported by good testing conclusion of the camera in question.
  
  I love a well constructed review. Since these cameras all use the same technology the difference is in the implimentation and that is hard to detect without good testing procedures.
  
  I personally have never seen a camera that does a good job on auto white balance in any conditions except outdoor natural sunlight. Some get close under tungsten but none get it right in my opinion.
  
  And I'm not a guy that tweaks and photopeeps my pictures. I load them into Photoshop Elements and at most fix color temperature, crop, and save.
  
  

• Wesley Fink - Thursday, April 3, 2008 - link

  Given the fact that we did use a 50mm f1.4 lens on all models (we even have a 25mm f1.4 Leica for the Olympus in reserve) your criticism that testing one camera with 1.4 and another with f4 would be acceptable is really unfair. We are glad that was brought to our attention and we have corrected it so all shots are at the same aperture.
  
  While the aperture initially varied in our test shots for ISO noise, that was never our intent. It should be obvious that if we went to the trouble of equipping all the DSLRs with f1.4 prime lenses that the testing was to be as close to apples to apples as we could make it with differing camera makers and different sensors. Those who pointed out the discrepancy in apertures were correct to do so and we reshot the whole series to correct our mistake of relying on the programs to set the same aperture line.
  
  However, i do not share the same enthusiasm for shooting all at f8. At 100 ISO and f4 the required shutter speed with our current lighting is about 2 seconds. To use a constant f8 the ISO 100 shutter speed would have to be around 8 seconds, and we would be running into noise issues that could be caused by long exposure times instead of ISO speed. You might say improve the lighting and shoot studio lighting, but that would defeat the purpose of looking at noise under demanding conditions.
  
  Most of our readers who read our camera reviews could care less how a camera might perform under studio conditions. They are more interested in seeing how it performs in conditions more like they would use it, and that is in an indoor setting with indoor type lighting where many of the worst DSLR pictures ever taken are shot. Our challenge is to find ways to scientifically test under those real-world conditions.
  
  It is not controlled conditions be damned, IMO, it is more how do you test controlled in the real world where these cameras will mainly be used by our readers?