Tentacle wrote:
I made the same mistake in expecting a considerable difference. Lets see...
300D vs 10D: 2.5 fps vs 3 fps, plastic vs magnesium, non wheel vs wheel. Same mpx. 10D had some more functionality, which the 300D could get in hacked firmware.
350D vs 20D/30D: 3 fps vs 5 fps plastic vs magnesium, non wheel vs wheel. Same mpx. Same thing with hacked firmware functionality. (Well, not the 30D spot metering or the 3 v 5 fps burst setting)
40D will have the same pixelcount, but do better with the scrollwheel, magnesium body, more fps and more processing speed (assuming DIGIC III is quite a bit faster than DIGIC II.)
Do you spot the trend here?
And this observation, ladies and gentlemen, is as apt as it is cynical ...Show more →
Nice observation.
I don't normally do this, but I would expect the following (in order of probability):
10.1MP
Anti-dust
DiG!C III
Possibly improved ISO (native 3200 ISO with 4800 ISO expansion)
Possibly improved DR (16bit)
5fps (possibly 6fps)
Possibly 9+6 AF points (same as 5D)
Possibly an anti glare coating on the LCD
Possibly a slightly improved vertical grip
Possibly a new shutter & mirror (quieter and longer life)
Possibly improved battery life
And that's about it. Metering, menus, software and functions all remain basically the same as 30D. I strongly doubt it will be 12MP, weather sealed, magnesium grip, 11+ AF points, up to 6400 ISO, etc.
And you know what? I'd trade up my pair of 20D's for a pair of those.
And I will venture out and do one more prediction, allthough this may also be archived under the header of "No Brainer":
- The 40D will still have the auto shutdown upon opening the CF compartment
As for jwil's predictions: Personally, I don't see the native ISO 3200 happening just yet, it will remain "extended". Maybe the 5D successor will have this, but I don't expect this on APS-C for now. It is possible, but not likely.
Same goes with Increased DR which is in fact even less likely since that requires bigger photodiodes. I only see that happen if Canon shoe-horns an APS-H sensor into the 40D.
I'm guessing that Digic3 will allow for some new tricks with noise control and DR and possibly allow for better performance. That's why I put them right after inclusion of D3, but other than than I agree that a DR improvement is unlikely. I think an ISO upgrade is about 50/50%
Go back some posts to my story on NR: There is physical signal NR, both for fixed pattern noise and for random pattern noise, which takes place in analogue signal space and then there is NR that's essentially software based, working with the discretised digital data. The first is mostly sensor-based, the latter is processing chip (read: DIGIC) based.
Sensor improvements will help reduce/prevent noise and, eventually, make true ISO 3200 possible. (Knowing Canon with their conservative ISO values, that'll actually be ISO 4000.)
If the processing chip causes the sensor to be underexposed by a stop and then pushes the digitised sensor output one stop, then it's false ISO 3200. DIGIC III has better (claimed) NR so I expect the 40D ISO3200 to look as clean as, or cleaner than, the 30D version. Still, that will be software routine based. DIGIC can't do anything on-sensor after all.
Same with DR: DIGICIII may be able to handle 16-bit RAW data, assuming that the A/D converters have at least that same bit-range, but if the sensor DR is the same amount of stops we had on the 20D and 30D sensors, then it will net only more graduations, not more range.
Having said that, more tonal graduations can still, potentially, increase image quality, even without increase in true DR. It will definately help with shadow noise reduction, given that dark areas (the far left on a histogram) inherently have less tonal range than highlights.
I understand all that, but what seems to be the case is you are assuming that there haven't been other improvements. I don't think that anyone can assume that at this point because there haven't been any serious test results posted of the 400D yet. Who knows at this point what may or may not have improved with that sensor or even if the sensor hitting the 40D will be the same one.
There are a lot of assumptions that Canon only alters one or two subsystems at a time for a revision, but for the 40D I'm betting on revisions to the sensor, processor, memory and buffer. That means advances in DR and ISO are not out of the realm of possibility. Probability is another issue.
I also think that many people assume that the current state of the art/science of how individual photodiodes work is all there is to it. Perhaps a change in the litho process, or wafer composition, or some other unknown advance can yield significantly lower noise or higher DR with a smaller photosite. How many would have predicted perpendicular storage on hard drives, DLP mirror arrays, or RISC computing would have lead to the breakthroughs in those respective areas.
And it seems Kodak is doing some interesting and ground breaking stuff with their newst sensors. I think assuming that a sensor MUST be manufactured the same way as older sensors and will have all the same limitations of older senors is seriously flawed logic.
Assume for a second that Canon is working on a higer resolution, higher DR, higer ISO chip for a new 1D. The basic research on this is probably done, and probably has been done for a while. But the research on how to make an individual photosite that could perform to spec for that type of FF high end sensor would have been done a very long time ago.
Now don't you think it's possible that Canon might have used that research as the basis for much of - if not all of - their new APS-C and APS-H sensors? From the very early review on Luminous Landscape, it appears that Canon has been able to increase resolution, decrease pixel pitch and maintain the noise levels of the old 350D:
But, my initial testing shows the XTi to be as noise free at most ISOs as its predecessor, so the extra resolution will be welcome. This size now allows nearly 16X20 " prints to be made, offering most photographers more than enough data for reasonably large prints, even with a fair amount of cropping.
Again, I think the anecdotal evidence is fairly stong that Canon may have done some significant work to the basic sensor design to allow for better performance, but the camera will need more processing power to take full advantage of it.
What I've seen over the years is that there is a continuous evolution going on. It's CMOS since the D30 and there have been gradual improvements in S/N ratio, pixel density and ... well. RAW has been 12-bit since the D30.
I have to chew on my own words too, because the 5D has a slightly lower DR than the 20/30D, despite having a bigger pixel size. OOPS. DR may improve with bigger photodiodes on paper (Canon claims so in their oft-mentioned FF white paper), it didn't do so for the 5D.
The thing is: There are advancements in S/N ratio despite decreasing pixel size. I don't think that Canon tackles one or two sub-problems at a time. More likely, Canon simply trades advancements in one area against another. SNR got better, but market/competition demands a higher pixel density. Result? Sacrifice some of the SNR improvements for more pixels. Noise per unit area is lower, but the pixels get smaller, so while the SNR improvement is still there, it is not as pronounced.
I forsee the same with the 40D: SNR improvement (be it both sensor-based, through design and production technology, and DIGIC-based, through software algorithms and more available processing power) part of it gets traded against the increased pixel density. Result: slightly better SNR, higher pixelcount. It's what we've seen since the D30.
Unless there is something REVOLUTIONARY about the 40D sensor this incremental evolution can be expected to continue. The revolution is not impossible per sé, but I would expect to see the alternative technology used on smaller sensors (digicompacts) before they go APS-C.
Tentacle wrote:
What I've seen over the years is that there is a continuous evolution going on. It's CMOS since the D30 and there have been gradual improvements in S/N ratio, pixel density and ... well. RAW has been 12-bit since the D30.
I have to chew on my own words too, because the 5D has a slightly lower DR than the 20/30D, despite having a bigger pixel size. OOPS. DR may improve with bigger photodiodes on paper (Canon claims so in their oft-mentioned FF white paper), it didn't do so for the 5D.
The thing is: There are advancements in S/N ratio despite decreasing pixel size. I don't think that Canon tackles one or two sub-problems at a time. More likely, Canon simply trades advancements in one area against another. SNR got better, but market/competition demands a higher pixel density. Result? Sacrifice some of the SNR improvements for more pixels. Noise per unit area is lower, but the pixels get smaller, so while the SNR improvement is still there, it is not as pronounced.
I forsee the same with the 40D: SNR improvement (be it both sensor-based, through design and production technology, and DIGIC-based, through software algorithms and more available processing power) part of it gets traded against the increased pixel density. Result: slightly better SNR, higher pixelcount. It's what we've seen since the D30.
Unless there is something REVOLUTIONARY about the 40D sensor this incremental evolution can be expected to continue. The revolution is not impossible per sé, but I would expect to see the alternative technology used on smaller sensors (digicompacts) before they go APS-C.
However, on the 400D, we saw that Canon improved pixel count and appears to have held noise constant WITHOUT upgrading the processor as it's still on Digic II. So, what does that mean if you improve another major system? Is there more untapped lattitude in the new sensor that more processing can unlock? Can the new chip actually go 3200ISO natively and get a software push to 4800? Did they cripple the DR and NR in software on the 400D to help differentiate it from a 40D?
Time will tell, but I'm betting we haven't seen the full potential from this new sensor yet. Probably a safe bet too, given that the sensor has only been on the market for 3 weeks.
I read more AF points; however, I'm more intrigued with faster AF vs. more AF points...
One would assume this 40D will be packed in the current 30D body and probably sell for around $1300 or so, but it still is not a sufficient backup body to the 1 series. Ideally the 40D specs with the 1D AF sensor packed inside a EOS 3 type body with optional VG would be the cats meow for me and I know for many others, I'd be willing to pay $1000 more for those features, something like that for around $2300 would be supreme. We can only hope that Canon get's wind of enough of these requests and finally succumbs to the pleas of the users
So...what are your comments on canon current line up: 18/9/2006, the 30D compared to the 400D?
Since they are 2 different generations of camera, it is fair to assume that the newer camera would have similar specs. The new generation of XXD should be on par with the XTi in pixel count, but there has always been a feature or 2 that simply sets them apart, be it AF points (XT vs. 20/30D) AI Servo (300D vs 10D), Fps (XT vs. 20/30D), Spot Meter etc. Layout of the camera with the scroll wheel and body composition has typically been the most visible of the differences, but imbeded in the camera, there too, have been differences.
Yes, better AF would definitely be the cat's meow. That's why I am excited to see the 400D with the 20D's AF - not just the location of the sensors but the SAME AF algorithms and enhanced central cross sensor. The rebel line has NEVER had the same AF capability as the xxD line, so I am hoping the tradition continues. The 300D did not have AI servo (hacked firmware DOES NOT COUNT), and the 350 did not have the enhanced central sensor.
It would be fantastic for everyone if the 40D had the 5D AF (including the 6 "hidden" central sensors) - because the 5D and most likely 1 series will get a boost as well.
jwil wrote:
However, on the 400D, we saw that Canon improved pixel count and appears to have held noise constant WITHOUT upgrading the processor as it's still on Digic II. So, what does that mean if you improve another major system? Is there more untapped lattitude in the new sensor that more processing can unlock? Can the new chip actually go 3200ISO natively and get a software push to 4800? Did they cripple the DR and NR in software on the 400D to help differentiate it from a 40D?
Time will tell, but I'm betting we haven't seen the full potential from this new sensor yet. Probably a safe bet too, given that the sensor has only been on the market for 3 weeks....Show more →
My guess is that the sensor on the 400D is designed as the 'last of the DIGICII' rather than the 'first of the DIGICIII', and that apart from the slightly wider microlenses it does not tell us much about what Canon are going to use in the 40D and so on.
FWIW my own guess is that we will just see a 1DsIII announced at Photokina, not scheduled for production until November, and that it will be a 22MP 8fps 16bit with no crop mode.
DR will not be significantly increased from the 1DsII in RAW, but JPEGs will do the equivalent of Sony's in-camera HDR.
Noise suppression of repetitive noise will be enhanced.
I would guess that other cameras will not be announced until the New Year, and they may have enhanced DR - edge issues at the dense FF make it too difficult to intro it on the 1DsIII.
Pure guesswork!
Regards,
DaveMart
timbop wrote:
...The 300D did not have AI servo (hacked firmware DOES NOT COUNT), and the 350 did not have the enhanced central sensor...
This brings up an interesting thought, I wonder if Canon will try and do something to prevent firmware hacks. I suppose this would depend on if Canon is ammused at the ingenuity of some who are able to accomplish this successfully, or if they are annoyed and want to see it end so people don't turn their 400D's into psuedo 40D's
However, on the 400D, we saw that Canon improved pixel count and appears to have held noise constant WITHOUT upgrading the processor as it's still on Digic II. So, what does that mean if you improve another major system? Is there more untapped lattitude in the new sensor that more processing can unlock?
I'm out on a limb here, since there is no hard info on what exactly it is that DIGIC does. I assume that Canon puts the A/D converters on the sensor because this saves on power. If that's accurate, then all that DIGIC can do is make a difference on the software side of NR.
Can the new chip actually go 3200ISO natively and get a software push to 4800? Did they cripple the DR and NR in software on the 400D to help differentiate it from a 40D?
No need to cripple DR and NR on the 400D. The DIGIC II is set to 12-bit RAW and has slower I/O to buffer and CF. Most likely its NR routine is not as advanced as DIGIC III. In other words, leaving the 400D with the DIGIC II is in itself effectively crippling.
Is it hypothetically possible that the sensor can do true ISO 3200 while the 400D stays limited to ISO 1600? Well, yes, it is possible, but what a waste of production resources that would be! Put extra effort into sensor production quality to get true ISO3200 and then not use it? Not likely, especially because the 400D will outsell the 40D. If Canon wants to give the 40D sensor an extra edge, it will simply use 2 different 10 mpixel sensors. It's what they did with the 350D/20D.
Time will tell, but I'm betting we haven't seen the full potential from this new sensor yet. Probably a safe bet too, given that the sensor has only been on the market for 3 weeks.
My money is either on A) some improvements because of better DIGIC III functions or B) considerable improvements because of DIGIC III and another sensor.
Tentacle wrote:
I have to chew on my own words too, because the 5D has a slightly lower DR than the 20/30D, despite having a bigger pixel size. OOPS. DR may improve with bigger photodiodes on paper (Canon claims so in their oft-mentioned FF white paper), it didn't do so for the 5D.
I've seen tests that show otherwise. Phil Askey's testing aside (he used out-of-camera JPEG's for the test, and the results were roughly equal), there are other resources that give varying amounts of difference presumably depending on the measurement criteria. Here's one that puts the 5D at about 1/3 stop ahead of the 20D:
The JPEG results from the Imatest test show similar performance to the 20D's JPEG (except at the highest quality standard where the 5D pulls ahead), though the 5D's RAW outperforms its JPEG by a fair margin. I must note that the difference between RAW and JPEG is not nearly as large as I had thought - I guess I need to caress my JPEG images a bit harder to see what I can get
There's a lot of talk as to whether or not the 40D will share the 400D sensor.
What part of the sensor are you actually measuring? The 400D will certainly have a 4 channel readout, compared to the 2 channels on the 400. This relevant metric alone makes it different. Assuming there isn't a major new sensor design waiting to be announced, you can bet your life the 40D will use identical technology with respect to gathering/processing photons. Simply because it's much cheaper to do so.
However, I'm expecting that true ISO3200 will be included (and probably true 1/3 ISO's too). Maybe a less agressive AA filter to boot. I'd also expect to see the 5D AF system simply because it doesn't make any sense NOT to put it in there. The 40D is aimed at a different market to the 5D (MkII?)
All the above is null and void if Canon are actually true to their word about 2007 being the year of greater DR.
What we really need is a better shutter design - can you imagine the marketing people talking about 1/16000 shutter speeds!!!!! (though that will need ISO128000...). It'll be almost as useful as a print button...hmmm
1/16000 shutters sound good for a number of people. shooting hummunging birds for one. bullets, for another, any place the requires a high speed image.
There is a market for everything.
Well, Minolta had a film camera with a top shutter speed of 1/12000.
So why not 1/16000?
I'm also wondering how much longer it will take until Canon will produce a DSLR with 10 fps, they've shown it's possible with the 1V. With Digic III and other technological advances I guess the read-out and data processing shouldn't be limiting anymore...
Edited by Philippe Arnez on Sep 19, 2006 at 12:34 AM GMT
Paul Gardner wrote:
1/16000 shutters sound good for a number of people.
For instance when you need to shoot @ F1,4 or F1.8 F2.8 for a nice DOF in the midle of a sunny day!
Now I've opened that can of worms, I think I'd rather go the other way - ISO25 (or less) would be extremely handy for longer exposures without resorting to NDs or high diffraction f-stops.
That's one thing that truely irritates me about the 20D: no ISO50.
grrrrrrrrrrrrr
I'd trade up from my 20D's simply to get: 10MP with the same noise and DR or 8MP with better DR, 50ISO, 9+6 point AF, 6-8fps, and a damn ISO setting indicator on the top display, in addition to what is already in the 30D.
