New t3i or new lens

[Daniel Browning]

Originally Posted by Taky Cheung

In that case, you won't get a darker picture.

I didn't say it would be darker. I said it would have the same intensity of light, but a much smaller area of light, and therefore more noise.

Originally Posted by Taky Cheung

There is no light loss.

Yes there is: all the light that was falling on the rest of the sensor, but is no longer used, is light lost.
That is the same reason why the R1 2K mode has more noise than 4K mode.

Originally Posted by Taky Cheung

The crop zoom has the same lens focal length, aperture size and everything. Same depth of field too. The difference is a narrower Field of View.

What do you mean by "The crop zoom"? Do you mean when the T3i is compared to itself in different modes? Or do you mean when the T3i in the 3X mode is compared to a 5D2? Do you agree with Pete Carney's position that T3i 3X with 50mm f/1.8 has the same depth of field as a 5D2 with 240mm f/1.8?

[Taky Cheung]

From what you said, the aperture value go from f/1.8 to f/8.6. That's huge amount of light loss causing the image being darker.

Let me put it this way. You took a picture with a 18 mega pixel camera. Use photoshop to resize it down to 1920x1080. Then, use the same picture, cut out the center 1920x1080 pixel. That's what this T3i 3x crop zoom is. There is just no lost of light.

Anyway, I don't have 5DM2 but I have this T3i camera that I have been using it at weddings. the feature works works great. There is no loss of light, depth of field, focal length... everything is the same, except there is a smaller FOV that makes it seems like a 3X zoom lens.

[Daniel Browning]

Originally Posted by Taky Cheung

From what you said, the aperture value go from f/1.8 to f/8.6.

No, I said f/1.8 on T3i 3X crop is *equivalent* to f/8.6 on the 5D2. Just like 50mm is *equivalent* to 240mm. Putting the lens on a T3i doesn't make the focal length go from 50mm to 240mm.

Originally Posted by Taky Cheung

That's huge amount of light loss causing the image being darker.

No, it's a huge amount of light loss causing the image to become *noisier*. Noise is a function of both light *intensity* and light *area*.

Let me illustrate it this way. Imagine three cameras:

• Medium Format camera with a 48x27mm sensor and 70 MP (pixel size 4.3 microns)
• Tiny digicam with 8.26x4.65mm sensor and 2.1 MP - 1920x1080 (pixel size 4.3 microns).
• Medium Format camera with a 48x27mm sensor in a "5.8X crop mode" - 2.1 MP - 1920x1080 (pixel size 4.3 microns).

Now, if you put the exact same 50mm f/1.8 lens on all three of these cameras, what do you get?

• Medium Format: wide angle of view, very low noise at ISO 1600.
• Tiny digicam: super telephoto AOV equivalent to 300mm on Medium Format, depth of field equivalent to f/10 on MF, and very high noise at ISO 1600
• Medium Format in 5.8X crop mode: super telephoto AOV equivalent to 300mm on Medium Format, depth of field equivalent to f/10 on MF, and very high noise at ISO 1600

Cropping is the *exact* same as using a smaller sensor in the first place. Now consider what kind of lens it would take to get the *same* image quality from all three:

• Medium Format: 290mm f/10.4 ISO 100
• Tiny digicam: 50mm f/1.8 ISO 3300
• Medium Format in 5.8X crop mode: 50mm f/1.8 ISO 3300

Again, the test I posted before demonstrates this (you can download the raw files or do your own DOF/noise tests to see for yourself).

None of this should come as a surprise. If it was possible to get the equivalent of a 800mm f/1.2 on Medium Format by using an 8mm f/1.2 on a mobile phone, then we'd all be doing that.

Originally Posted by Taky Cheung

Let me put it this way. You took a picture with a 18 mega pixel camera. Use photoshop to resize it down to 1920x1080. Then, use the same picture, cut out the center 1920x1080 pixel. That's what this T3i 3x crop zoom is.

Exactly!

Originally Posted by Taky Cheung

There is just no lost of light.

If there is no loss of light, then explain why doing the exact steps you described results in the cropped version having much more noise than the downsized version?

Originally Posted by Taky Cheung

everything is the same, except there is a smaller FOV that makes it seems like a 3X zoom lens.

Everything is *not* the same.

[Taky Cheung]

I didn't bother to read all your comparison or illustration on with using another camera..(no offense =) ha). I already did my own test with my T3i. That's all I concern is the output from my T3i. I do not see any quality loss, light loss. It makes perfect resulting footage that I can use at weddings. Giving me more zoom power during shooting rather than taking time to switch lens.. That's what's more important to me than those theoretical illustrations..

You would think cropping means using a smaller sensor in the first place. However, if the image is already captured in full then cropped to 1920x1080, I don't see why would it cause quality loss.

Everything is the same. You are still shooting with the same camera, same lens with the same focal length, same scene, same DOF. The only difference is less FOV giving an illusion of a digital zoom.

I'm kinda tired of doing this again. You can read this thread. It's the same thing we are talking about
http://www.dvinfo.net/forum/canon-eo...parison-3.html

[Keith Betters]

I've been reading this thread with great interest. I have no dog in the fight, and all I can do is explain what my eyes are seeing. I somewhat follow the technical talk but I don't understand all of it. When I am cutting between my t2i's and my t3i in digital zoom mode I SEE A DIFFERENCE. Not a big one but there is a difference. Trust me, I don't want it to be. This is one of the reasons I bought the t3i so I could use the digital zoom feature. However, I think it falls slightly short. For me right now it will have to work as I can not afford a 70-200 lens of any kind.

As far as someone on here mentioning the glass that you put in front it, I do not agree. Reason being because if you are using the same glass on the other cameras but not noticing any quality fall off, I would attribute that to the feature and not the lens.

I'm not saying the quality is not good, and that it can not be used, but what I'm saying is I see a difference when using the feature and when not. Especially when they are all filming the same thing at the same time!

[Daniel Browning]

Originally Posted by Taky Cheung

I'm kinda tired of doing this again. You can read this thread. It's the same thing we are talking about
http://www.dvinfo.net/forum/canon-eo...parison-3.html

I did read the thread. It's full of the same misconceptions, myths, and urban legends that you've repeated here. What's your point?

[Murray Christian]

Originally Posted by Keith Betters

As far as someone on here mentioning the glass that you put in front it, I do not agree. Reason being because if you are using the same glass on the other cameras but not noticing any quality fall off, I would attribute that to the feature and not the lens.

But the feature is effectively zooming in on a small part of the image, and therefore lens, where it wasn't before.
I think that was the point. For instance, the Canon 18-55 kit lens is quite weak in the middle at various points on its zoom. If you shoot video with it, because its downscaled from the sensor, it's not all that noticeable to most people. If you use a 3x crop of the same image off the sensor you are really going to see that softness now, where an A grade lens wouldn't show the same problem.

This doesn't negate your experience. The mode may do all sorts of things. I've never seen it. But I think the point was that this mode can highlight weaknesses in a lot of lenses that you wouldn't normally see (and perhaps other things besides).

[John Wiley]

Originally Posted by Keith Betters

As far as someone on here mentioning the glass that you put in front it, I do not agree. Reason being because if you are using the same glass on the other cameras but not noticing any quality fall off, I would attribute that to the feature and not the lens.

The glass does matter, because when you use crop mode you are seeing an expanded view of the lens. And like anything to do with imaging, when you expand it, flaws become more apparent. It's like looking at the lens through a magnifying glass - everything gets bigger, including the problems.

It is the same as the examples they show in lens reviews to demonstrate flaws. They show a full photo and it looks fine. Then they show a 1:1 crop (which is what 3x crop mode is) and suddenly you can notice all sorts of softness and chromatic abberation. Take a look at a few lens reviews and the 1:1 crops and you'll see that many lenses are not up to the task in 1:1 crop mode, but they are perfectly acceptable for regular 1920x1080 video, or even 18mp photographs (as long as they are not displayed too large), because they are using the whole image circle.

[Nigel Barker]

Daniel, you are comparing raw still photographs to prove your point whereas what you should be comparing is 1920x1080 video shot on the 5DII & from the 600D/T3i. The 5DII introduces its own artefacts (particularly moire & aliasing) when downsampling in camera from the full sensor 5616x3744 pixels to 1920x1080 pixels for HD video whereas the 600D/T3i is remarkably free of these artefacts as it does no in camera downsampling but records from a central 1920x1080 pixel area of the sensor. So even though the 600D/T3i 3X HD crop image may be noisier it will also be 'cleaner' in other aspects.

[Daniel Browning]

Originally Posted by Nigel Barker

So even though the 600D/T3i 3X HD crop image may be noisier it will also be 'cleaner' in other aspects.

Agreed, but that is separate from and in addition to the effect of cropping, and we should treat it as such. Even if the normal video mode was so poor that the crop mode was actually able to match it on noise (and it isn't), we'd still do well to recognize the two different factors that go into making it so, rather than believing incorrectly that the crop mode is always just as good.

[Kin Lau]

Originally Posted by Daniel Browning

Unfortunately, that's not correct. In every way that matters, the 50mm f/1.8 on the T3i at 3X is equivalent to a 240mm f/8.6, not f/1.8. That's not just for depth of field, either. They have the same *intensity* of light (since both are f/1.8), but light gathering (noise) depends on the intensity *plus* area. And since a real 240mm f/1.8 and 5D2 has way more area, it would also have significantly less noise. For example, you would get the same noise with T3i 3X, f/1.8, 1/60, ISO 100 as you would from the 5D2, 240mm, f/8.6, ISO 1600. I've demonstrated the principle here:

Image comparison

There is no such thing as a free lunch.

This comparison is _very_ misleading, and I would go so far to say that it's just wrong when you're discussing the 3x crop on a T3i. Actually, it's also wrong for any sort of video comparison.

You're comparing images re-sampled from a full-frame 21mp RAW image instead of the 1920x1080 capture that happens during HD. This way of re-sampling will reduce noise, but that's not what's used in the Canon HD capture process.

Why we get aliasing, is that pixel binning is not happening, we only get line skipping, so collecting the pixels from every 3rd line vs only at the center of the sensor will make no difference as to sensor noise.

In summary, the 5Dm2 does not resample 21mp into 2mp for a 1920x1080 image, it line skips, so you only ever get about 1920x1080 pixels worth of info. Cropping the sensor still only gives you the same amount of info, same basic number of pixels.

[James Donnelly]

Originally Posted by Daniel Browning

Unfortunately, that's not correct. In every way that matters, the 50mm f/1.8 on the T3i at 3X is equivalent to a 240mm f/8.6, not f/1.8. That's not just for depth of field, either. They have the same *intensity* of light (since both are f/1.8), but light gathering (noise) depends on the intensity *plus* area. And since a real 240mm f/1.8 and 5D2 has way more area, it would also have significantly less noise. For example, you would get the same noise with T3i 3X, f/1.8, 1/60, ISO 100 as you would from the 5D2, 240mm, f/8.6, ISO 1600. I've demonstrated the principle here:

Image comparison

There is no such thing as a free lunch.

Sorry, but I can't see how your discourse "sheds any light" on the discussion about 600d cropping. Could you explain how that is related to the 600x crop feature for video?

Could you explain why does noise depends on area? I thought it depended on pixel density.

If you shine a flash light into a 600d so that each pixel receieves equal light intensity, what difference does it make which pixels are sampled to produce your 1080p image?

If you are talking about noise comparing the full sensor to the crop zone, then yes, but you're not. The crop simply changes which pixels are recruited to produce the image, so it shouldn't make a difference. No offence, but every time I read what you have written I get the impression you maybe don't appreciate how pixel binning and line skipping is used to generate the 1080p image in the first place. That's how I see it. Please help me understand if I'm wrong.

Any difference in image quality could explained by the fact that the sensor physically does not have a 3x 1920 width, so 3x is necessarily a slight reduction in resolution, no?

Why don't you write another article comparing frames from cropped and non-cropped 600d footage? I think it would help to settle this issue.

The only effect I can see being an issue is that you will have a DOF which is derived from the effective new sensor size during crop mode, ie; not as shallow as non-cropped focal length equivalent FOV.

Happy to be proven wrong, but please be nice :)

Update: just seen Kin Lau's post, which explains my point of view much more succinctly!

[Ger Griffin]

ok this is what I have.
I have 2 t3is and a 7d
I have 2 canon 17-55mm 2.8
and I have a canon 70-200 2.8 IS II
I also have a canon 135mm f2

The 70-200 is on the 7d.
When I take it out and use it the images blow any other combination out of the water.

There is no replacement for fantastic glass. the two t3is with a 17-55 2.8 crop gets the job done. they are my main work cameras. The other heavy combination tends to stay in the bag most of the day. But when I do take it out, shoot with it, and get the footage back home, I promise myself that next time Im going to use it more. Because yes, it is BY FAR a nicer looking picture.

BTW if you shoot in 3x keep that iso down below I would say 640.
3x is good to have when needed. But should be avoided if possible.
For example, I would opt for 55 mm in 1x over 17mm in 3x mode.
Similar perceived framing, but different in quality. And this is for more
reasons that just DOF. Noise, sharpness, CA (basically everything)

But dont take offence if you exclusively shoot 3x (its still amazing compared to what we were getting before)
Its when mixing lenses with cameras as I do regularly the IQ difference is more obvious during post.

These are my own real world observations.

[Daniel Browning]

Originally Posted by Kin Lau

Actually, it's also wrong for any sort of video comparison.

I disagree. The primary difference is that video does more of the processing in-camera, but other than that the comparison is quite similar.

Originally Posted by Kin Lau

In summary, the 5Dm2 does not resample 21mp into 2mp for a 1920x1080 image, it line skips, so you only ever get about 1920x1080 pixels worth of info. Cropping the sensor still only gives you the same amount of info, same basic number of pixels.

I don't it can really be summarized as just "about 1920x1080 pixels worth of info" -- it's too complicated for that. In any case, but as I mentioned in the last post, the losses from poor sampling and resampling are separate from and in addition to the effects from cropping.

[Daniel Browning]

Originally Posted by James Donnelly

Could you explain why does noise depends on area? I thought it depended on pixel density.

Most photography and video with any camera made in the last 8 years or so is photon-shot-noise limited. So the amount of noise depends solely and entirely on the total number of photons detected by the image sensor. Modern sensors show the same quantum efficiency (QE, or percentage of light detected) between a huge variety of pixels sizes: from 80 square microns all the way to 3 sq. um. Many would have thought this was not possible because a relatively larger amount of area is lost to the metal between the photodiodes, but it turns out that gapless microlenses and other innovations have overcome that entirely. So whether you collect 50000 photons in 2 million large pixels or 5000 photons in 20 million small pixels, the total amount of light detected (and therefore the noise power from photon shot noise at any given spatial frequency) remains the same.

Originally Posted by James Donnelly

Why don't you write another article comparing frames from cropped and non-cropped 600d footage? I think it would help to settle this issue.

Thanks for the suggestion. I don't have the 600D.

Originally Posted by James Donnelly

The only effect I can see being an issue is that you will have a DOF which is derived from the effective new sensor size during crop mode, ie; not as shallow as non-cropped focal length equivalent FOV.

I agree about the DOF.