T4i- affordable Digic 5 DSLR?

* thanks, very interesting
* bummer: I was afraid the sensor would be a bottleneck too... 3x rolling shutter is definitely too much, 60ms would be crazy (only useful if we can control when to use line-skipping and when to avoid moire)
* let's put our hopes on the 24mpix new-sensor rumors
 
Sorry to switch gears, but with all this technical talk, does anyone know why the 600D can do 3-10x digital zoom, but not 1-3x? And could this be overcome by a lack of line-skipping? Because frankly, a DSLR that could do, say, 1-5x digital zoom and maintain true 1080p res would be a real force to be reckoned with, especially if you're using a lens that doesn't suffer from a lot of CA. Imagine being able to slap a 1.4/24 on your cam and be able to shoot the equivalent of 24-120.
 
Last edited:
Technically all digital zoom poses quality lost as it's just multiplying pixels. 600D at 3X is not really digital zoom. It is using the inner 1920x1080 pixels out of the 18 megapixel sensor.
 
Taky said:
Technically all digital zoom poses quality lost as it's just multiplying pixels. 600D at 3X is not really digital zoom. It is using the inner 1920x1080 pixels out of the 18 megapixel sensor.
Click to expand...

Well, we can play semantics with the term "digital zoom" all we like; that's what Canon calls it, regardless of the fact that it's far superior to the resampling nonsense still cameras have. What I'm wondering is if 1-3x will be theoretically made possible on the next generation.
 
yes, when the 600D/T3i came out, I was also wondering why no 1x-3x digital zoom

I guess it has to do with what Rick explained about sensor readout times: if you read more than 1/3 of the lines then rolling shutter kills your images; still, they could have gone for "read the central 2/3 of the lines, skipping 1/2 of them", which would have yielded 1.5x digital zoom (but still with aliasing/moire issues, though probably less than in standard mode); maybe not worth the effort to code that (in which case, I'd just be happy if they gave a1ex and the rest of the ML developers the relevant info so that they could code all this stuff themselves :) )
 
NormanBates said:
yes, when the 600D/T3i came out, I was also wondering why no 1x-3x digital zoom

I guess it has to do with what Rick explained about sensor readout times: if you read more than 1/3 of the lines then rolling shutter kills your images; still, they could have gone for "read the central 2/3 of the lines, skipping 1/2 of them", which would have yielded 1.5x digital zoom (but still with aliasing/moire issues, though probably less than in standard mode); maybe not worth the effort to code that (in which case, I'd just be happy if they gave a1ex and the rest of the ML developers the relevant info so that they could code all this stuff themselves :) )
Click to expand...

I think it also has to do with the fact that the current method is, 3x is just a "setting" you choose, and then it just does cropping from then on until 10x with no real resolution change, the way it does with traditional "digital zoom." Cropping requires very little processing power compared to what I'd imagine a true resampling of a different segment of the sensor at 1080p would take. But frankly, even if there wasn't a zoom, if they'd just allow a selectable option of say, 1x, 2x, 3x, 4x, 5x, even without a smooth zoom throughout the range, that'd be excellent, and it would make great primes much more versatile. I guess you could call it "digital tele conversion"?
 
I just searched for some samples of S100 footage on youtube, and cried tears. The days of my trusty 550D are numbered.
She will be put on side with the 85mm on for closeups. That's called being retired :)


Even thinking about getting the S100 and a Gorillapod for 2nd unit while waiting for the Canon 660D!
 
Thanks for that link NB!!! The Sony NEX 5N sure does do well in all the tests vs. the GH2 and Canon's. Most importantly it has 1080-60p. I may have to look into one of those if the new Canon doesn't have that mode. It is super important for me for show motion shots. The Canon's 720-60p is so lame it's not worth every using for my purposes.

Cheers,
Pete
 
c3hammer said:
Thanks for that link NB!!! The Sony NEX 5N sure does do well in all the tests vs. the GH2 and Canon's. Most importantly it has 1080-60p. I may have to look into one of those if the new Canon doesn't have that mode. It is super important for me for show motion shots. The Canon's 720-60p is so lame it's not worth every using for my purposes.

Cheers,
Pete
Click to expand...

I really want to like the NEX-5N, especially because of the specs, size, and the fact that you can use more old MF lenses (like Rokkor & FD), but something about the footage I've seen just seems very...tape-y? It doesn't look suitable for films the way the Canons do. I'd love to be wrong about this, though; the ability to grab a NEX-5N (or the upcoming NEX-7) and an Atomos Ninja would just be a killer replacement for the T3i.
 
beware:
* the biggest problem I see with the 5N is the overheating (even faster than the 7D, it seems)
* I think the nex-7, A65 and A77 have a lot more noise than the 5N; I've seen some footage, and some negative comments, but no good side by side comparison
* the link I posted from DXOmark only looks at the stills mode; when translating that to the video mode, you'll have to think how the lack of a CineStyle picture style affects the 5N, and how the line-skipping increases noise on the canon DSLRs (or, just better, look for info elsewhere; for the S100, though, there's not a lot of tests around yet, and I'm not sure there will be any really good ones)
 
abe in space said:
I really want to like the NEX-5N, especially because of the specs, size, and the fact that you can use more old MF lenses (like Rokkor & FD), but something about the footage I've seen just seems very...tape-y? It doesn't look suitable for films the way the Canons do. I'd love to be wrong about this, though; the ability to grab a NEX-5N (or the upcoming NEX-7) and an Atomos Ninja would just be a killer replacement for the T3i.
Click to expand...
I don't know about tapy-y, but it is a different look. Average joe home video is what it is, but with proper lenes, composition and lighting it looks simply stunning. I think this one from Philip Bloom fits the latter catagory :)

http://vimeo.com/31064478

Cheers,
Pete
 
Was really hoping to hear something about this from CES today. So far, all we got is the new G1x (which is awesome for photography fans as a no-compromise point&shoot camera).

Sigh.
 
Bother.

I had some hopes for this camera, given the I-Frame movie in its spec.

However.

The sensor is very definately *not* a DSLR shape (At least not Canon DSLR) so there is still hope for the new 650D.

Remember the diferance between Digic V and digic V+, I suspect we'll only see those in single-digit cameras, the rest wil have to make do with the Digic V
 
Poop. That is disappointing.

Well... This is a long shot, but I'm wondering if they just used their old routines for that camera, because it is not a DSLR. Think about it- The camera EOSHD is discussing is a bit of an anomaly. Since it is designed to be a point and shoot, they probably kept the software and algorithms from the P&S line. But we've already seen what the does to larger sensor cameras, as evidenced by the current crop of Canon HDSLRs. The P&S cameras never needed a newer algorithm because their sensor does not exhibit the same artifact problems (as mentioned earlier in this thread, even the Digic4 point and shoot cameras had relatively clean video- the artifact problems we see appear to be from the combination of large sensor + line skipping algorithm).

Canon is aware of the moire/aliasing problems on DSLR cameras (as evidenced by their press releases around the 1Dx). It could very well be that the DSLR firmware will treat video encoding differently, but they did not use DSLR-grade software for the G1x.

Since the P&S cameras were fine the way they processed video before, it is possible that they didn't see a need to update the video processing system in models of this class.

Since the G1x happens to have a larger sensor, we're seeing the same problems as before. Hopefully, the 650D, 5DmIII, etc, will be different.

...or, it is also likely that Canon doesn't care, and would rather draw more attention to their video-centric cameras instead. Too soon to tell.
 
M. Gilden said:
I'm wondering if they just used their old routines for that camera, because it is not a DSLR. Think about it- The camera EOSHD is discussing is a bit of an anomaly. Since it is designed to be a point and shoot, they probably kept the software and algorithms from the P&S line.
Click to expand...

I want to believe this also. However the samples from the Canon S100 P&S actually look cleaner and more detailed than the G1X.

It looks to me like a case of product segmentation, and I can only hope that what they are segregating the G1X from is the next bunch of Canon DSLRs, and not something like the C300. Because that makes no sense at all. If the 5D Mark III performs like the DIGIC 5 does in the G1X, it is all over for Canon in the DSLR video market.
 
Andrew Reid said:
I want to believe this also. However the samples from the Canon S100 P&S actually look cleaner and more detailed than the G1X.
Click to expand...

It was pointed out earlier in the thread that regular point and shoot models from Canon don't have the same moire and aliasing problems. So it shouldn't be any surprise that the s100 doesn't. The G1x is the only one with the problem, as far as I can tell. Even the Digic4 models don't suffer from this.
So why did the DSLR models have the problem?

I'm guessing it has to do with the resolution and overall size of the sensor. The P&S cameras have less resolution to begin with, which means they are throwing out less data when line skipping. Combine that with the fact that the photo cells are physically smaller and closer together, and you have a reasonably clean image with a simple line-skipping downsample.

But make those photo cells bigger, farther apart, and a larger number of them, and suddenly you have a recipe for disaster.

The G1x is different because despite being a point and shoot, it has a sensor almost the size of APS-C (closer to a m4/3 sensor, if I recall correctly, right?). That means you're running into DSLR-sized problems.

But again, since this is technically a P&S, and the other models are find resolving video this way, why change that bit of software for this one single camera? If Canon has shown me anything, its that they like to keep things consistent across devices in the same class of product. For example, when the "lowly" rebel t2i came out, I don't think any of us expected the video IQ to be on par with the powerhouse 7D at the time. Keeping things consistant like that cuts manufacturing and R&D costs.
While the G1x is a really interesting stills camera, it is priced at the point that some people would prefer a DSLR rebel instead. And those who don't will want a cheaper P&S. The G1x has really caught the interest of a number of pro photographers who want a better quality P&S to carry around, but that is an admittedly very small demographic. Possibly one that Canon doesn't think will sell enough models to make so many changes to their classic P&S firmware for. Like I said before, this camera could really just be an anomaly.

If the 5D Mark III performs like the DIGIC 5 does in the G1X, it is all over for Canon in the DSLR video market.
Click to expand...

Well, we could use more full-frame sensors, to be honest. But I agree, Canon needs to step up their game and show us what the Digic5 is capable of. We know its fast enough to do it, and we know Canon is aware of the problems. Its really up to them to decide to make it happen, or continue using the same line-skipping as their P&S line, despite documented problems with it.
 
You must log in or register to reply here.
Back
Top