Reading mode: Light Dark

Very confused on cropping!! Related to HS10

Go to last post
Mister T said:
Sorry, I did say I'd had a quick glance, and yes your answer is right at just under 1MP, and yes it is hard to write the squared bit into the equation and make it clear - I've just tried several ways and the clearest I could manage was:

(4368 / 3.8) x (2912 / 3.8) = 1150 x 766
Sorry for being so pedantic ;)
Click to expand...
I agree that's better (and correct). But as the saying goes "same difference" LOL! I actually thought of expressing it like that but wanted to group (4368x2912) together so it is better visually .
Mister T said:
Not sure many HS10 users would agree that the HS10 Image should be resized from 10MP to 1MP for the comparison! Yes I know that's fine for most prints but I don't think many pixel peepers will agree ... (ducks his head and runs for the trenches) ...
Click to expand...
Well according to the calculations the DSLR would need to be a crop of 1150x766 or 1MP to match a 720mm FOV and it can't be any bigger than that. So if comparing IQ then the HS10 must also be downsized to match.
Mister T said:
I would suggest a compromise resolution of, say 3 or 4MP, for comparison.
Click to expand...
Yes that's possible but only if you use a much longer lens for the DSLR. One must use a 300mm equiv FL to compare at 2MP and a 400mm equiv FL to compare at 4MP. Full HD is 1920x1080 or 2MP and it is a good common ground for IQ comparison.
Mister T said:
The interpolated DSLR image might look soft but, in my experience, a bit of PP in Photoshop and you can easily convince people that the image is actually of a higher resolution (in the same manner that all digital cameras kid us by having, for example, 10M light cell receptors but really only 2.5M pixels of data - as each pixel uses a red, a blue and 2 green cells - if my understanding is correct) .
Click to expand...
I've never seen any great outcome on any upsized images regardless of any camera it came from, even from the most expensive and best software out there.

...
 
RoysLaw said:
I've never seen any great outcome on any upsized images regardless of any camera it came from, even from the most expensive and best software out there.
Click to expand...
Once upon a day I would have said this but look at this 100% crop from a 6mp sensor upsized to 12mp (from a camcorder!) - bottom picture.

http://forums.dpreview.com/forums/read.asp?forum=1012&message=35270652

... and as I might have mentioned before, doesn't a 10mp sensor have 2.5m red cells, 2.5m blue cells and 5m green cells so really that's only 2.5m pixels? So up-scaling is a well practised and mature science that fools us all?
 
maisatomai said:
HS10 gives you a image 3,648 x 2,736 and focal length is 720mm. The lens on HS10 is actually a 126mm f5.6 lens.

Imagine if this lens can be taken off HS10 and put into a 5D. It will give a 4368 x 2912 and focal length is 126mm. We can crop it using PC to emulate the 720mm image that HS10 give. To do that, we need to divide 4368 X 2912 by 5.7. We will get 766 X 510. Won't this dimension not even meet the minimum recommended dimension to print a 6X4 photo?

I understand the theory that a lens will always be a lens, no matter which camera you use it. What am I missing out here to complete my understanding?
Click to expand...
If you have a look at the web page linked to below, you will see a comprehensive “template” that can be used to help solve problems of the nature you describe:

http://www.robsphotography.co.nz/crop-factor-advantage-appendix-3.html

The above web page compares two “theoretical” cameras, so that the various relationships between pixel density, pixel size and image size, can be seen clearly for both linear and area relationships.

You could apply the formulas in the above page to help answer your question. But it helps you to understand the principles if you first work through a theoretical example that is not complicated by the roundings and imprecise mathematical relationships that can occur with "real world" examples.

For a practical application of the above "template", have a look at this one, which compares the full frame Sony A900 with the newly announced APS-C Sony NEX-3 and NEX-5.

http://www.robsphotography.co.nz/Sony-nex-A900.html

Regards
Rob
 
Mister T said:
... and as I might have mentioned before, doesn't a 10mp sensor have 2.5m red cells, 2.5m blue cells and 5m green cells so really that's only 2.5m pixels? So up-scaling is a well practised and mature science that fools us all?
Click to expand...
Pixels are primarily defined spatially, not in terms of color capability. I have an old Macintosh with a 512x384 screen where each pixel can be either black or white; yet they are still pixels.

A 10MP bayer sensor does indeed have a photosite breakdown by color as you outline, but it does capture light at 10M spatially distinct points, and so it's not a stretch to call it 10 megapixels. It cannot resolve color differences at the full resolution, but can (nearly) resolve luminance differences at full resolution, and the eye tends to be more sensitive to luminance details than chromatic details. (Many JPEG files store color information at a lower resolution than luminance information for this very reason.)

With a Bayer sensor, it is of course necessary to interpolate the color information (called "demosaicing") in producing a usable image. This isn't the same as upscaling, since there is some image data to guide the process.

There are sensors that capture all three color channels at each photosite, most notably the foveon sensors used in Sigma cameras. A 4.8ish MP foveon sensor (generally advertized as 14 MP by Sigma, but that's another story) seems to match somewhere around an 8-10MP bayer sensor in practical resolution, and some of that is undoubtedly due to the omitted antialias filter, which causes aliasing artifacts at times. The technology does have some significant drawbacks--most notably, it lacks sensitivity (or, said another way, is noiser at high ISO) when compared with bayer sensors, and the color accuracy is not so great because the color channels exhibit a fair amount of crosstalk.
--
--DrewE
 
DrewE said:
Mister T said:
... and as I might have mentioned before, doesn't a 10mp sensor have 2.5m red cells, 2.5m blue cells and 5m green cells so really that's only 2.5m pixels? So up-scaling is a well practised and mature science that fools us all?
Click to expand...
Pixels are primarily defined spatially, not in terms of color capability. I have an old Macintosh with a 512x384 screen where each pixel can be either black or white; yet they are still pixels.

A 10MP bayer sensor does indeed have a photosite breakdown by color as you outline, but it does capture light at 10M spatially distinct points, and so it's not a stretch to call it 10 megapixels. It cannot resolve color differences at the full resolution, but can (nearly) resolve luminance differences at full resolution, and the eye tends to be more sensitive to luminance details than chromatic details. (Many JPEG files store color information at a lower resolution than luminance information for this very reason.)

With a Bayer sensor, it is of course necessary to interpolate the color information (called "demosaicing") in producing a usable image. This isn't the same as upscaling, since there is some image data to guide the process.
Click to expand...
Point taken, should have noticed this. This is much the same way that colour information is put in the PAL TV signal (I don't know if NTSC is the same) where the 6MHz bandwidth, originally designed just for luminance in black and white days, has the colour signal 'hidden' in it but at a lower resolution.

Is it fair to say that the colour resolution is lower and that areas of similar brightness with fine colour patterns will suffer a bit of 'mushiness' (good technical term there)?
 
After reading such insightful explaination, I have another question: His choosing of the APS-C camera and FF.

For FF, he chooses 24 megapixels camera. For APS-C, he chooses 10.67 megapixels camera. Both camera actually has same pixel density (10.67*1.5*1.5=24 megapixels). Nowadays, 550D megapixel is 18MP. 18*1.5*1.5=40.5 megapixel). 1D mark iv is only 16.1 megapixel so can I say that crop factor is useful after all?

In HS10, the crop factor is 5.62x. Using the same formula as above, to compare HS10 with a FF, the FF will need (10.3*5.62*5.62=325.31932 megapixels). To put it in another way, the pixel density is much more in HS10.

Thus, can I say "to really be sure a cropped version of a 200mm lens picture quality is much better than HS10, we must need a 325.31932MP FF DSLR). Or else, we cannot say all that"

Actually, I was hoping that some one can do some sort of a field test to see whether HS10 can match with a 200mm lens. Although there been claims by ppl that HS10 is beaten by 200mm lens easily, I was thinking could it be due to the human error involved. For example, the person who made this claim is with respect with his birding experience. With a 200mm lens, it is so easy to handhold. But with a HS10 at 30X, it is impossible to handhold.
maisatomai said:
Is this website yours?

I love this article most! http://www.robsphotography.co.nz/crop-factor-advantage.html

Great site!!!
Click to expand...
 
maisatomai said:
Thus, can I say "to really be sure a cropped version of a 200mm lens picture quality is much better than HS10, we must need a 325.31932MP FF DSLR). Or else, we cannot say all that"

Actually, I was hoping that some one can do some sort of a field test to see whether HS10 can match with a 200mm lens. Although there been claims by ppl that HS10 is beaten by 200mm lens easily, I was thinking could it be due to the human error involved.
Click to expand...
It is not beaten easily ... the HS10 holds its own. I found some instances early on where I could show that they were beaten easily, but that was probably down to operator error.

So in my new articles, I shot them both head to head.

http://kimletkeman.blogspot.com/2010/05/hs10-vs-d300-vs-f70exr-test-of-reach.html

http://kimletkeman.blogspot.com/2010/05/hs10-vs-d300-vs-f70exr-test-of-reach_16.html

--
http://kimletkeman.blogspot.com
http://letkeman.net/Photos
 
Thanks for doing the experiment. Actually from the leaf photo, I prefer HS10 shot. I seriously cannot tell that it is leaf behind the branches for D300. For HS10, the color of the leaf is green and nice.

Not bad at all for a $500 HS10 against a heavier $2300 D300+lens, even before taking into consideration HS10 can take Full HD video and D300 can't.

Of course, there are many limitation of HS10.

For example, if Letkeman has done a comparsion of shooting at wide angle, I think D300 has beaten the hell out of HS10.
If Letkeman has shot in low light, HS10 would also been beaten totally.
 
Kim Letkeman said:
So in my new articles, I shot them both head to head.

http://kimletkeman.blogspot.com/2010/05/hs10-vs-d300-vs-f70exr-test-of-reach.html

http://kimletkeman.blogspot.com/2010/05/hs10-vs-d300-vs-f70exr-test-of-reach_16.html
Click to expand...
Thanks for the tests - quite interesting.

To be honest I was surprised how well the HS10 did hold up under this scrutiny and I'd agree that overall it was a tie (in this specific test case).

I did note that in the licence plate detail you could nearly read the text at the bottom of the plate (??? TO DISCOVER) although we know it has a major resolution advantage in this scenario.

I noted in a photo (from at least 10m away) of my daughter in the 'band' at a school show that her face looked a little mushy but the pianist's music was nearly clear enough to read. Below is a 100% crop OOC. The detail on the music was surprisingly good IMHO and shows that just occasionally the HS10 gets somewhere near its 10mp resolution? Well, in black and white anyway ;)



 
Mister T said:
I did note that in the licence plate detail you could nearly read the text at the bottom of the plate (??? TO DISCOVER) although we know it has a major resolution advantage in this scenario.
Click to expand...
Yours To Discover ... Ontario's motto (an enormous province with many different eco systems etc.)
Mister T said:
I noted in a photo (from at least 10m away) of my daughter in the 'band' at a school show that her face looked a little mushy but the pianist's music was nearly clear enough to read. Below is a 100% crop OOC. The detail on the music was surprisingly good IMHO and shows that just occasionally the HS10 gets somewhere near its 10mp resolution? Well, in black and white anyway ;)
Click to expand...
Cool image, and shows just how well the HS10 resolves relatively high contrast detail from a distance.

It's Achilles heal is that distant foliage and grass and feathers tend to be low contrast details and that's a different story unfortunately.
Mister T said:
Click to expand...
--
http://kimletkeman.blogspot.com
http://letkeman.net/Photos
 
maisatomai said:
After reading such insightful explaination, I have another question: His choosing of the APS-C camera and FF.

For FF, he chooses 24 megapixels camera. For APS-C, he chooses 10.67 megapixels camera. Both camera actually has same pixel density (10.67*1.5*1.5=24 megapixels). Nowadays, 550D megapixel is 18MP. 18*1.5*1.5=40.5 megapixel). 1D mark iv is only 16.1 megapixel so can I say that crop factor is useful after all?

In HS10, the crop factor is 5.62x. Using the same formula as above, to compare HS10 with a FF, the FF will need (10.3*5.62*5.62=325.31932 megapixels). To put it in another way, the pixel density is much more in HS10.

Thus, can I say "to really be sure a cropped version of a 200mm lens picture quality is much better than HS10, we must need a 325.31932MP FF DSLR). Or else, we cannot say all that"

Actually, I was hoping that some one can do some sort of a field test to see whether HS10 can match with a 200mm lens. Although there been claims by ppl that HS10 is beaten by 200mm lens easily, I was thinking could it be due to the human error involved. For example, the person who made this claim is with respect with his birding experience. With a 200mm lens, it is so easy to handhold. But with a HS10 at 30X, it is impossible to handhold.
maisatomai said:
Is this website yours?

I love this article most! http://www.robsphotography.co.nz/crop-factor-advantage.html

Great site!!!
Click to expand...
Click to expand...
Yes, this is my hobby web site, thanks very much for your kind comments.

I haven’t had any hands-on experience with the HS10, so I had a look at this review:

http://www.photoreview.com.au/reviews/advanced/fujifilm-finepix-hs10.aspx

I see that the HS10 uses a 4.2mm – 126mm lens, which in 35mm terms, provides a focal length range of about 24mm – 720mm.

When you use the “template” I have on my site here:

http://www.robsphotography.co.nz/crop-factor-advantage-appendix-3.html

the crop factor is calculated by dividing the sensor width of a full frame camera (36mm) by the sensor width of the second camera, which in the case of the HX10 is shown in the above article as 6.16mm.

So, using my template, for the arithmetic to “reconcile” properly, the crop factor would be regarded as about 5.84 (36mm divided by 6.16mm).

However, when analysing “real world” examples, there are always “roundings” to contend with, and I guess this is why you say the HS10 has a crop factor of 5.62x?

It’s interesting that the full frame Canon 5D has image dimensions of 4368 pixels x 2912 pixels on a sensor with a size of 36mm x 24mm. In contrast, the Fujifilm HS10 has 3648 pixels x 2736 pixels on a sensor with a size of 6.16mm x 4.62mm.

So, the pixel density (in megapixels per square centimetre) of the 5D is about 1.5, while the pixel density of the HS10 is about 36 megapixels per square centimetre. Therefore, if the Canon 5D had the same pixel density as the HS10, it would have about 300 megapixels.

Regards
Rob
http://www.robsphotography.co.nz
 
Robsphoto said:
So, using my template, for the arithmetic to “reconcile” properly, the crop factor would be regarded as about 5.84 (36mm divided by 6.16mm).

However, when analysing “real world” examples, there are always “roundings” to contend with, and I guess this is why you say the HS10 has a crop factor of 5.62x?
Click to expand...
And to add further confusion, the crop factor is most useful when it reflects the actual area of the sensor that is used. That way, multiplying the focal length by the crop factor gives an accurate effective focal length.

In this case, Fuji report an actual focal length of 126mm and an effective focal length of 720mm.

That gives a crop factor of 5.7 ... splitting the difference between both of your results.

--
http://kimletkeman.blogspot.com
http://letkeman.net/Photos
 
Kim Letkeman said:
tdkd13 said:
Not sure where you have been seeing 35mm equivelant focal lengths but it probably wouldnt be on an actual lens. Even the bridge and point and shoot cameras print the actual focal length on the lens. In this picture of an S100fs it says "f=7.1-101.5mm" right on the lens, that is the true focal length. Below that is a Canon SX200 which says "5.0 - 60mm" again, that is the true focallength.
Click to expand...
And , the information isn't useful at all for common comparison unless you know the crop factor ...
Click to expand...
Thanks for that,

I'm from the era of 35mm SLR when interchangeable lenses didn't lead to confusion - I've not seriously looked at DSLRs (although I have used the odd one!) so I'd always assumed focal lengths would be quoted in 35mm equivalents as has been the norm on P&S since I got my Fuji F2800 (sold as "equivalent to a 38-228mm zoom on a 35mm camera" in 2002).

Yes I know the lens shows the actual focal range, even on a basic P&S, but they always sell the camera quoting the equivalent.

At least I'm now aware of crop factor if I'm ever rash enough to go back to (d)SLRs - particularly relevant at the wide angle end.

P.s. One of the most user friendly things IMO about the HS10 is the use of both scales on the lens. Obviously ( facepalm ) you can't do that with DSLR lenses because crop factors change.
 
Mister T said:
P.s. One of the most user friendly things IMO about the HS10 is the use of both scales on the lens. Obviously ( facepalm ) you can't do that with DSLR lenses because crop factors change.
Click to expand...
Hello

Please is there any simpler way I understand what you talking about?

--
Best regards
JCardoso
 
jcardoso said:
Mister T said:
P.s. One of the most user friendly things IMO about the HS10 is the use of both scales on the lens. Obviously ( facepalm ) you can't do that with DSLR lenses because crop factors change.
Click to expand...
Please is there any simpler way I understand what you talking about?
Click to expand...
Everything we have today is "normalized" to focal lengths that everyone has been familiar with for many years, that being the focal lengths that cover the sensor on a 35mm camera -- i.e. a full frame sensor.

That's where the term "35mm equivalence" comes from. You will also see it as "135 equivalence" sometimes, 135 being the film format designation for those cameras.

So ... the focal lengths in those days were what you often see quoted. 28mm was the beginning of wide, 20mm was really wide. 50mm was "normal" ... about what you see just standing there. 135mm was a common mid telephoto and 200mm a nice long telephoto.

Now ... on a sensor like the Fuji F70EXR, focal ranges are from 5mm to 50mm, which translates to 27,, to 270mm in 35mm equivalence.

The crop factor is simply a multiplier that takes you from the smaller sensor size or focal length to the 35mm equivalent.

In this case, the crop factor is achieved by dividing the 35mm equiv by the lens focal length ... 270/50 == 5.4 ... that's a moderate crop factor. The HS10 is closer to 5.7 ... the S100fs closer to 4. An APS-C sensor is 1.5.

So what Mister T is saying is that it would be nice to print the real focal lengths on lenses and their 35mm equivalents ...

Ultimately, though, it does not matter as all interchangeable lenses for dSLRs are already normalized to full frame focal lengths, even those that don't cast the right image circle. It's weird, but there it is.

--
http://kimletkeman.blogspot.com
http://letkeman.net/Photos
 
You must log in or register to reply here.

Similar threads

Yannis1976
Discussion about sensor and lenses for nature and birds...
2 3
Replies
45
Views
3K
dazzerfong
D
Yannis1976
Starting a discussion on sensor size and lenses for nature and birding...
Replies
14
Views
538
baobob
baobob
A
Fuj X100VI vs m43 or Sony rangefinders: can you help me understand equivalence and dimensions?
2
Replies
31
Views
948
alcelc
alcelc
S
  • Question Question
Super telephoto and teleconverter vs digiscoping on a spotting scope for video
Replies
10
Views
641
lehill
lehill
X
Digital Back with technical camera crop factor confusion for 35mm equivalent
Replies
4
Views
25
ampimagedotcom
ampimagedotcom

Keyboard shortcuts

f
Forum
m
My threads
Mobile site
About
Editorial content
Cameras & Lenses
Community
All content, design, and layout are Copyright © 1998–2025 Digital Photography Review All Rights Reserved.
Reproduction in whole or part in any form or medium without specific written permission is prohibited.
When you use DPReview links to buy products, the site may earn a commission.
©GPS Media - Guides, Products, Services.
Back
Top