SX50: thoughts on digital zoom

Started Jun 17, 2013 | Discussions
Don_Campbell Senior Member • Posts: 2,709
SX50: thoughts on digital zoom

Digital zoom is a quick and handy way to get even more magnification that full optical zoom with these amazing cameras--that's an obvious "given." How much do you give up in IQ using digital zoom was a question that ran through my mind yesterday as I spent several hours watching big, beautiful birds sit still and do nothing more interesting than scratch their heads or ruffle their feathers--while I waited for interesting behavior.

I decided to to a test: take an image shot raw and then processed for a good final image, crop it and then upscale it 4x to compare with a full digital zoom image taken of the same scene at the same time of day, same exposure and same ISO.

Immediately there are all the quibbles anyone can dream up: how can you say that this is a fair comparison if you get to tweak it from the raw through the final JPG and the DZ photo doesn't get that treatment? Well, that's sort of the issue isn't it? When the camera does it the camera limits what you get by applying its algorithms not the ones available to me and my software. So--no comparison is "fair" by some measures but if the question is what is given up just accepting the DZ image it seems fair enough.

First: full digitial zoom to "200%" of the falcon of some of my other photos:

Digital zoom image.

Next, a raw file converted by dpp with slight contrast adjustment, then gently processed for a little more NR in raw therapee. That image was then cropped to 1000 x 750 px and upscaled to 4000x3000 using GIMP's sinc (Lanczos3) scaling and finally slight sharpening using GIMP's "smart sharpening" plugin. It sounds more complicated than it is in practice. It took about 2 minutes to get it to GIMP and another minute or two to crop, upscale and sharpen.

Upscaled image converted from raw and tweaked.

The answer (for me) is that yes, I can do better by performing my own "digital zoom" outside the camera. Alternatively, the camera's DZ image is pretty good and it's ready to go as soon as it is taken, which is a clear convenience.

There is one way that DZ is very much superior--it actually allows you to see in the viewfinder objects that are so far away that you can't really tell what is happening if you are not digitally zoomed. I've generally "looked" by taking a photo and then magnifying in the LCD display and DZ is way more convenient and faster than that.

Thanks for looking.

Don

Augustin Man
Augustin Man Veteran Member • Posts: 7,432
Re: SX50: thoughts on digital zoom

I agree with you: the pp-ed RAW image is slightly better and if magnified with remarkable less noise. Didn't you try the equivalent Tc (2X)? I'm asking you because in my tests the Tc 2X provided a better image than the equivalent DZ...

Thank you for this useful comparison!

Happy shooting,

Augustin

 Augustin Man's gear list:Augustin Man's gear list
Canon PowerShot SX50 HS Nikon Coolpix P900
gardenersassistant Veteran Member • Posts: 3,849
Re: SX50: thoughts on digital zoom

An interesting comparison and conclusions. Thanks for taking the time to document your experiment and conclusions.

I have been working along somewhat similar lines with my SX240. I am now using CHDK to capture RAW + JPEG.

I use digital zoom for framing shots beyond the 20x optical limit of the SX240 (500mm equivalent focal length), but then do the crop myself from the RAW version, which covers the 20x optical area irrespective of the amount of digital zoom. This means, like you point out, being able to see what is going on better, especially as you move further beyond the optical limit. Also, the precise framing of the image is very important for my visual preferences (usually involving flowers when using the SX240), and using digital zoom lets me see how the framing will look rather than having to visualise it from the larger-coverage (optical limit) image (which frankly my imagination/visual capabilities are not up to).

Also, when going for an image beyond the optical limit, and working hand-held (sometimes one-handed for shots at awkward angles) as I do with the SX240, there is a lot of movement of the image on the LCD. I can't hand-hold steadily enough to keep the framing exactly where I want it. However, when I am in digital zoom territory it doesn't matter if I slightly cut off one side or another in various attempts at the shot, because there is "slack" available around the edges that I can make use of when doing the crop later.

Doing the crop later has another advantage. I have the JPEG to remind me what I had in mind when I took the shot, but more than once I have discovered, when looking at the RAW version with its wider coverage, that there is a crop that I like better which uses more of the RAW image than what I saw on the LCD at shooting time. Of course if I had actually used the digital zoom JPEG as the source of my image I wouldn't be able to do that.

And there is a big benefit in handling noise. I tend to use rather high ISOs, but details and textures matter a lot to me. The higher ISO JPEGs have lots of detail/texture smudged away irretrievably by the in-camera noise reduction (and there is no option on the SX240 for turning the noise reduction down). Processing a RAW version means I get to decide what compromise to make between noise reduction and detail retention, suitable for the size of the output medium and my own preferences as to the "look" of my images. I can also use more subtle, image-specific approaches to noise reduction than the camera, for example applying different levels of noise reduction to the subject and the background, and/or to different parts of the subject and/or different parts of the background.

RAW also gives benefits in terms of retaining details in highlight areas/recovering areas that are blown in JPEG versions despite my habitual under-exposure to try to avoid it.

All in all I'm sure this is the best way for me to work, but I have to say in the couple of days after I got the camera and before I started using RAW, I had been pleasantly surprised at the image quality that the camera could produce up to 2x or perhaps slightly more digital zoom (ie actual, camera-produced JPEG digital zoom rather than "DIY digital zoom"/cropping from RAW that I've been talking about here).

(unknown member) Veteran Member • Posts: 3,073
Re: SX50: thoughts on digital zoom

Don, that digital zoom photo looks pretty darn good to me!  What a nice additional weapon to have in a pinch.  Thanks for sharing your photos and dialog.

Murry

-- hide signature --
OP Don_Campbell Senior Member • Posts: 2,709
Re: SX50: thoughts on digital zoom

Augustin,

Thanks for the comment.

Augustin Man wrote:

I agree with you: the pp-ed RAW image is slightly better and if magnified with remarkable less noise. Didn't you try the equivalent Tc (2X)? I'm asking you because in my tests the Tc 2X provided a better image than the equivalent DZ...

No, I didn't test the fixed Tc versions of DZ mostly because I was interested in dealing with the extremes of zoom and the 2X Tc is only half the zoom of the DZ at its maximum.

The strategy of the DZ seems curious to me: crop and then upscale the image to the original sensor's full size seems kind of asking for trouble. More typical for me is to be too distant to fully zoom into the subject but to get my version of a well framed photo using cropping with little or no upscaling.

Thank you for this useful comparison!

Happy shooting,

Augustin

Don

OP Don_Campbell Senior Member • Posts: 2,709
Re: SX50: thoughts on digital zoom

gardenersassistant wrote:

An interesting comparison and conclusions. Thanks for taking the time to document your experiment and conclusions.

I have been working along somewhat similar lines with my SX240. I am now using CHDK to capture RAW + JPEG.

I use digital zoom for framing shots beyond the 20x optical limit of the SX240 (500mm equivalent focal length), but then do the crop myself from the RAW version, which covers the 20x optical area irrespective of the amount of digital zoom. This means, like you point out, being able to see what is going on better, especially as you move further beyond the optical limit. Also, the precise framing of the image is very important for my visual preferences (usually involving flowers when using the SX240), and using digital zoom lets me see how the framing will look rather than having to visualise it from the larger-coverage (optical limit) image (which frankly my imagination/visual capabilities are not up to).

A very interesting difference between shooting raw natively (which disables digital zoom) and shooting it with CHDK (which doesn't). I am certain to try this with CHDK in the SX50. I never thought to use CHDK that way with my SX20 which I shot almost exclusively with raw after CHDK came out for it.

Also, when going for an image beyond the optical limit, and working hand-held (sometimes one-handed for shots at awkward angles) as I do with the SX240, there is a lot of movement of the image on the LCD. I can't hand-hold steadily enough to keep the framing exactly where I want it. However, when I am in digital zoom territory it doesn't matter if I slightly cut off one side or another in various attempts at the shot, because there is "slack" available around the edges that I can make use of when doing the crop later.

Another good comment and idea!

Doing the crop later has another advantage. I have the JPEG to remind me what I had in mind when I took the shot, but more than once I have discovered, when looking at the RAW version with its wider coverage, that there is a crop that I like better which uses more of the RAW image than what I saw on the LCD at shooting time. Of course if I had actually used the digital zoom JPEG as the source of my image I wouldn't be able to do that.

And there is a big benefit in handling noise. I tend to use rather high ISOs, but details and textures matter a lot to me. The higher ISO JPEGs have lots of detail/texture smudged away irretrievably by the in-camera noise reduction (and there is no option on the SX240 for turning the noise reduction down). Processing a RAW version means I get to decide what compromise to make between noise reduction and detail retention, suitable for the size of the output medium and my own preferences as to the "look" of my images. I can also use more subtle, image-specific approaches to noise reduction than the camera, for example applying different levels of noise reduction to the subject and the background, and/or to different parts of the subject and/or different parts of the background.

That's basically why I used CHDK almost exclusively for raw with my SX20.

RAW also gives benefits in terms of retaining details in highlight areas/recovering areas that are blown in JPEG versions despite my habitual under-exposure to try to avoid it.

All in all I'm sure this is the best way for me to work, but I have to say in the couple of days after I got the camera and before I started using RAW, I had been pleasantly surprised at the image quality that the camera could produce up to 2x or perhaps slightly more digital zoom (ie actual, camera-produced JPEG digital zoom rather than "DIY digital zoom"/cropping from RAW that I've been talking about here).

Thanks for your thoughtful comments and analysis.

Don

Mr. Click Senior Member • Posts: 1,105
Re: SX50: thoughts on digital zoom

Here's something I've noticed, that I haven't seen anyone else comment on.  You can get more accurate exposure, in some situations, using the digital zoom.  Say you have a contrasty scene, and your subject is only a quarter of the frame at 1200mm.  Without manual intervention, the brightness other three-quarters of the frame will control the exposure the camera chooses.  By zooming in to 4800mm, your subject will fill the frame, and its exposure will be be closer to correct.  You can somewhat see it in your example, because the whites are better exposed in the zoomed-in version than in the cropped version.

-- hide signature --

Photograph—a 1/125 second slice of time that is able to store decades of memories.

 Mr. Click's gear list:Mr. Click's gear list
Nikon D500
Britney Elvis Veteran Member • Posts: 5,072
Hey Mr Click

Mr. Click wrote:

Here's something I've noticed, that I haven't seen anyone else comment on. You can get more accurate exposure, in some situations, using the digital zoom. Say you have a contrasty scene, and your subject is only a quarter of the frame at 1200mm. Without manual intervention, the brightness other three-quarters of the frame will control the exposure the camera chooses. By zooming in to 4800mm, your subject will fill the frame, and its exposure will be be closer to correct. You can somewhat see it in your example, because the whites are better exposed in the zoomed-in version than in the cropped version.

-- hide signature --

Photograph—a 1/125 second slice of time that is able to store decades of memories.

I wondered about that. But dismissed it, thinking that the evaluative metering can really only be using existing photocells on the chip for its algorithm, and they like the Raw data... are pre-digital zoom. But I could be completely wrong.

I do know that with my dslr and Manual Focus lenses, the digital zoom did allow me, what I believed to be some more accurate focus.
gus

-- hide signature --

"It is not necessary to understand things in order to argue about them."
~~ Pierre Beaumarchais~~

OP Don_Campbell Senior Member • Posts: 2,709
Re: SX50: thoughts on digital zoom

Mr. Click wrote:

Here's something I've noticed, that I haven't seen anyone else comment on. You can get more accurate exposure, in some situations, using the digital zoom. Say you have a contrasty scene, and your subject is only a quarter of the frame at 1200mm. Without manual intervention, the brightness other three-quarters of the frame will control the exposure the camera chooses. By zooming in to 4800mm, your subject will fill the frame, and its exposure will be be closer to correct. You can somewhat see it in your example, because the whites are better exposed in the zoomed-in version than in the cropped version.

Thanks for your comment.

I suppose it is possible that exposure is affected by DZ but I don't think my images demonstrate that. The exposures of the two images are the same. There may be a slight difference introduced by the contrast tweak of my pp of the raw but I think it is far more influenced by the shadow of the bird's beak that reduces the bright white spot on the DZ'd image and is placed differently in the non-DZ'd image.

Don

-- hide signature --

Photograph—a 1/125 second slice of time that is able to store decades of memories.

GeraldW Veteran Member • Posts: 7,791
Re: SX50: thoughts on digital zoom

Don,

Nice experiment.  There's a lot of advice in print to avoid using DZ because it degrades the image.  However, I've disregarded that and for years have used DZ, starting with a G5 back in 2004.  I've found that Canon does a pretty good job of implementing DZ on their cameras, in spite of up-sampling the image back to full resolution.  4x is a lot of DZ, and I usually don't go higher than 2x to avoid the artifacts introduced by up-sampling.

Panasonic, with their Extra Optical Zoom (EZ) crops out the center of the image like everyone else; but they don't up-sample back to full resolution to avoid the artifacts.  Not much difference at small increases; but a bit better at 2x DZ.

In your test, you PP'd the RAW image - a huge advantage.  Why not PP the JPEG?  With the right editor, it works very well; although not all editors work equally well on JPEGs - and some very badly.

Years ago, I experimented with DZ Vs ISO.  You can get away with a lot more DZ at lower ISO settings.  That experiment was repeated several times with a Canon G5, a G7, and a Pro 1, all Canon cameras and so all using upsampling.  Printed at 8" x 10", I saw no difference in image quality, until the file size dropped below 3 MP (before upsampling, of course) at ISO 100.  Based on that criteria, a camera like your SX50HS with 12 MP, should be able to use more DZ than my 5 MP G5.  If you only view on-screen, or print at 4" x 6", you can use even more DZ, so long as the ISO doesn't get too high.  I have since found that a sharp 3 MP file prints very nicely as a borderless 8.5" x 11".

Since Digital Zoom is just a form of cropping, the guidelines I use also apply to crops in PP, and by extension, to the maximum print size a given crop with print at.

-- hide signature --

Jerry

 GeraldW's gear list:GeraldW's gear list
Canon PowerShot S90 Panasonic Lumix DMC-ZS7 Canon PowerShot ELPH 330 HS Canon PowerShot SX260 HS Panasonic Lumix DMC-FZ200 +2 more
OP Don_Campbell Senior Member • Posts: 2,709
Re: SX50: thoughts on digital zoom

GeraldW wrote:

Don,

Nice experiment. There's a lot of advice in print to avoid using DZ because it degrades the image. However, I've disregarded that and for years have used DZ, starting with a G5 back in 2004. I've found that Canon does a pretty good job of implementing DZ on their cameras, in spite of up-sampling the image back to full resolution. 4x is a lot of DZ, and I usually don't go higher than 2x to avoid the artifacts introduced by up-sampling.

Panasonic, with their Extra Optical Zoom (EZ) crops out the center of the image like everyone else; but they don't up-sample back to full resolution to avoid the artifacts. Not much difference at small increases; but a bit better at 2x DZ.

In your test, you PP'd the RAW image - a huge advantage. Why not PP the JPEG?

Thanks for the comments. I didn't PP the JPEG because I was comparing what one could do with processing the raw compared to what Canon does processing the image to JPEG.

I recommend reading the interesting post by gardenersassistant above to see the possibilities of using DZ for framing but getting a full sized raw (DNG actually) via CHDK to process carefully for better results. I certainly plan to check that out myself.

With the right editor, it works very well; although not all editors work equally well on JPEGs - and some very badly.

I've been pp-ing all kinds of images for quite a while. The basic problem with OOC jpgs is that the camera's processing has already applied NR that often kills detail that you cannot recover under any circumstance. In the case of superzooms, the challenge is always holding down noise while retaining detail and using the best of image editors for each part of the pp. Upsampling the size of the image is just an added image-damaging process since the smaller image will generally do a better job than the upsampled one in printing or Internet displaying.

Years ago, I experimented with DZ Vs ISO. You can get away with a lot more DZ at lower ISO settings. That experiment was repeated several times with a Canon G5, a G7, and a Pro 1, all Canon cameras and so all using upsampling. Printed at 8" x 10", I saw no difference in image quality, until the file size dropped below 3 MP (before upsampling, of course) at ISO 100. Based on that criteria, a camera like your SX50HS with 12 MP, should be able to use more DZ than my 5 MP G5. If you only view on-screen, or print at 4" x 6", you can use even more DZ, so long as the ISO doesn't get too high. I have since found that a sharp 3 MP file prints very nicely as a borderless 8.5" x 11".

Since Digital Zoom is just a form of cropping,

cropping plus processing plus upscaling...yes a form of cropping with some degree of degrading the image in the full process. I have no experience testing any possible enhancement of printing that might come from upscaling but I have my doubts.

the guidelines I use also apply to crops in PP, and by extension, to the maximum print size a given crop with print at.

-- hide signature --

Jerry

Thanks for your detailed comments.

Don

I2K4 Contributing Member • Posts: 526
Re: SX50: thoughts on digital zoom

Interesting topic. Using CHDK RAW/DNG will not affect func/set adjustments for the companion JPEG images, which can be an advantage over native Canon RAW + JPEG that disables them. My SX10 pre-dates the in-camera noise reduction, something I'm often grateful for when I see smeary "fingerpainted" output from NR - if forced to high ISO I'd rather DIY the noise reduction.

On the main topic, I've generally cropped instead of digital zooming, using CHDK DNG since I know I'm going to go into post-processing anyway. But I do recall some online commentary back when I was figuring out the camera, that indicated digital zooming can be better in some situations, where exposure or focus settings, for example, would be materially different for the close-in final composition, and could be thrown off by the wider angle optical shot you want to crop. The mechanics of it are hazy in my memory, but I've kept that back of mind when I'm pretty sure I'll want to crop significantly.  Could be a question of spot metering or changing depth of field for that wide angle optical shot.

It may be worth shooting both (though a wildlife shot is often a one-off proposition.)

 I2K4's gear list:I2K4's gear list
Canon PowerShot S5 IS Canon PowerShot SX10 IS Canon PowerShot SD4000 IS +1 more
gardenersassistant Veteran Member • Posts: 3,849
Re: SX50: thoughts on digital zoom

Don_Campbell wrote:

Mr. Click wrote:

Here's something I've noticed, that I haven't seen anyone else comment on. You can get more accurate exposure, in some situations, using the digital zoom. Say you have a contrasty scene, and your subject is only a quarter of the frame at 1200mm. Without manual intervention, the brightness other three-quarters of the frame will control the exposure the camera chooses. By zooming in to 4800mm, your subject will fill the frame, and its exposure will be be closer to correct. You can somewhat see it in your example, because the whites are better exposed in the zoomed-in version than in the cropped version.

Thanks for your comment.

I suppose it is possible that exposure is affected by DZ but I don't think my images demonstrate that. The exposures of the two images are the same. There may be a slight difference introduced by the contrast tweak of my pp of the raw but I think it is far more influenced by the shadow of the bird's beak that reduces the bright white spot on the DZ'd image and is placed differently in the non-DZ'd image.

Here is an example. On the left are the DNG and JPEG from a single exposure, using evaluative metering, and on the right the DNG and JPEG from another exposure, also using evaluative metering. The one on the right used full optical zoom. The one on the left used 4x digital zoom.

The camera was on a tripod. The DNG versions are framed the same, but the exposures differ by one stop. (To check this wasn't an artefact of the tree moving in the breeze I zoomed back and forth several times from optical zoom only to 4x digital zoom and got consistent differences in exposure each time). This is consistent with the camera doing its metering based on (or at least influenced by) the area that will be in the JPEG (ie after any digital zoom crop).

OP Don_Campbell Senior Member • Posts: 2,709
Re: SX50: thoughts on digital zoom

Thanks for the interesting and illustrative example. I wouldn't be surprised to find that it depended on the type of metering selected, but that would take more tests.

Regards, Don

gardenersassistant Veteran Member • Posts: 3,849
Re: SX50: thoughts on digital zoom

Don_Campbell wrote:

I wouldn't be surprised to find that it depended on the type of metering selected, but that would take more tests.

I used evaluative metering because I assumed that the effect would not apply for centre-weighted or spot metering. However ...

... I just did a test. I was wrong.

Here we have a pair of captures, on the left using 4x digital zoom, on the right using full optical zoom only. This pair of captures used centre-weighted metering. I did this several times and although the breeze was causing a lot of movement in the trees the results were consistent. Here too, there is a one stop difference. I was a bit surprised by this, but not totally dumbstruck - after all it is weighting the metering towards the centre, not basing the metering on the centre of the image.

I could not draw any conclusion about spot metering. With the breeze keeping the foliage in constant movement, the metering changed between readings. For 4x digital zoom it varied (over 20 or so measurements) from 1/160 to 1/500. For a similar number of measurements using only full optical zoom it varied from 1/60 to 1/320. This might look like an average one stop or so difference, but it is "the wrong way round" (the digital zoom measurements giving faster shutter speeds rather than, with the other two tests, and understandably so in the context, slower), so I suspect what I saw with spot metering was just a statistical fluctuation.

Keyboard shortcuts:
FForum MMy threads