The Camera Sensor - Impact and Utilization

[Pica]

Hello Everyone,
Background: I've been photographing for several years, and like to stop at times and contemplate both technical and photographic matters. This question concerns the first.

I have a 16 Mega-pixel sensor which I of course use mostly as a 16 MP sensor. However I also have the possibility of running the camera sensor in e.g. 12 MP mode. My technical question is: If one decides to change the mode from 16 to 12 MP, will this change the overall quality on photographs, e.g. the dynamic range, or is it just that the technical implementation for example blocks a number of pixels to achieve the result of 12 Mega-pixels?


Thank you for your cooperation!

[Dave Humphries]

I think you need to tell us which brand and model camera you are discussing Pica, so we can research how they are doing it and whether it has other effects.

Thanks, Dave

[Pica]

Hello again Dave,
Thank you for your response.

With a background as a systems engineer and project leader in technique projects, I have been hoping for that the camera vendors had reached a consensus on how to implement the feature; or at least that one could see a rule of thumb! But from your reply I interpret that this is not the case. Is it so? (Is it more like the none generic raw-strategy? No general technical implementation?)

Yes of course I will supply the data you are asking for. I am now looking for a new camera and thus I choose:
- DSLR: Nikon D5300
- Bridge: Fujifilm FinePix HS50 EXR

Looking forward to your result of the research you propose.

All the Best.
Pica

I think you need to tell us which brand and model camera you are discussing Pica, so we can research how they are doing it and whether it has other effects.

Thanks, Dave

[rpcrowe]

This is not an answer to your question but, a question regarding your query. At the price of memory these days, why would anyone want to shoot in any other than full size??? I cannot, for the life of me think of any valid reason...

[xpatUSA]

This is not an answer to your question but, a question regarding your query. At the price of memory these days, why would anyone want to shoot in any other than full size??? I cannot, for the life of me think of any valid reason...

For me, full size brings slow processing, even on a 4-cpu Core i5 machine, and a image size 4704x3136px that I will never use because I don't print.

So, I shoot in LO res raw (2x2 binned) 2336x1658px.

As for Pica's concerns, as far as I know, the D5300 produces it's lower image sizes by re-sampling - unlike my Sigma which uses true binning on-sensor.

http://www.dpreview.com/reviews/nikon-d5300/2

If I were concerned about D5300 image IQ, I would shoot full-size and use the re-sampling method of my choice, not Nikon's.

[george013]

Pica,

I'm not sure I understand you. The D5300 is a crop camera. It doesn't have the possibility to use a part of the sensor. Different JPG sizes are due to compression. And more compression means less quality. Full frame camera's sometime have the possibility to use only a part of the sensor so you get a DX format and can use DX lenses.
I don't know of the Fujifilm.

George

[Black Pearl]

Pica,

I'm not sure I understand you. The D5300 is a crop camera. It doesn't have the possibility to use a part of the sensor. Different JPG sizes are due to compression. And more compression means less quality. Full frame camera's sometime have the possibility to use only a part of the sensor so you get a DX format and can use DX lenses.
I don't know of the Fujifilm.

George

Yes it does - as does pretty much every DSLR I have ever come across regardless of sensor size.

In the case of the D5300 you can set it to; Large 6000x4000, Medium 4496 x 3000 or Small 2992 x 2000
These sizes are irrespective of the jpeg compression settings which can be set to Fine or Normal.

[george013]

Yes it does - as does pretty much every DSLR I have ever come across regardless of sensor size.

In the case of the D5300 you can set it to; Large 6000x4000, Medium 4496 x 3000 or Small 2992 x 2000
These sizes are irrespective of the jpeg compression settings which can be set to Fine or Normal.

The frame doesn't change when I'm right. The pixels are getting "bigger". You're right, it's not pure jpg compression, it's another form of compression. The sensor is used for 100%. Quality is getting less.
It's easier to talk about the raw, then you know what the camera does, and not something as pp.

George

ps
I think down sampling is the technique.

[dje]

Pica

The specs for the Fuji HS50 show the following

"still image
L : (4:3) 4608 x 3456 / (3:2) 4608 x 3072 / (16:9) 4608 x 2592 / (1:1) 3456 x 3456
M : (4:3) 3264 x 2448 / (3:2) 3264 x 2176 / (16:9) 3264 x 1840 / (1:1) 2432 x 2432
S : (4:3) 2304 x 1728 / (3:2) 2304 x 1536 / (16:9) 1920 x 1080 / (1:1) 1728 x 1728"

Shooting in L mode, with an aspect ratio of 4:3 all 16M pixels are being used. With 3:2 and 16:9 aspect ratios,some of the top and bottom of the frame is chopped off. These aspect ratios use 14MP and 12 MP respectively. In these modes, the image quality should be no different to the full 16MP. It's just that some of the pixels aren't used.

With M and S modes, there is downsizing occurring and there will some loss in image quality.

Dave

[Manfred M]

Pica - let's think about this in terms of the optics of what can and does happen.

1. The sensor has a fixed size and the lens use will be designed to have an image circle that provides coverage to the full sensor. This tells me that for any final image size change, unless we see the "multiplier effect" that we see in crop frame cameras (which we don't see here), the full sensor is being used to capture the image.

2. With a DSLR and the optical viewfinder, it is difficult to implement variable aspect ratios. The Fuji, being a bridge camera with an electronic viewfinder / display, this is relatively easy. We just add black bars to crop the display to the desired aspect ratio. This is no different than the way our computer screens or televisions display videos shot at different aspect ratios.

3. The best way to understand what is happening inside the camera is to see what is happening with the raw data. I did this with both my Nikon D800 (full-frame) and my Panasonic GX7, a micro two-thirds crop frame (2x factor).

With the Nikon, I got smaller jpeg files when I changed the image size setting, but the raw files stayed the same size, regardless. With the Panasonic, I saw a similar effect with image sizes (jpegs changed dependent of the image size settings, but the raw files stayed constant), but when I changed the aspect ratio, the raw files changed size (slightly), depending on the aspect ratio selected, so that suggests to me that there is an electronic crop occurring on the raw data, dependent on the selected aspect ratio).

So, my conclusion is that the smaller size settings perform a similar function to how a jpeg would perform when we resize with when we edit on our computers with post-processing software (i.e downsampling). The raw files are not affected at all by this setting, except when the aspect ratio is changed. Here there is a small decrease in size as fewer pixels are captured, but this stays the same at a given aspect ratio, regardless of the image size selected.

[Saorsa]

Pica

The specs for the Fuji HS50 show the following

"still image
L : (4:3) 4608 x 3456 / (3:2) 4608 x 3072 / (16:9) 4608 x 2592 / (1:1) 3456 x 3456
M : (4:3) 3264 x 2448 / (3:2) 3264 x 2176 / (16:9) 3264 x 1840 / (1:1) 2432 x 2432
S : (4:3) 2304 x 1728 / (3:2) 2304 x 1536 / (16:9) 1920 x 1080 / (1:1) 1728 x 1728"

Shooting in L mode, with an aspect ratio of 4:3 all 16M pixels are being used. With 3:2 and 16:9 aspect ratios,some of the top and bottom of the frame is chopped off. These aspect ratios use 14MP and 12 MP respectively. In these modes, the image quality should be no different to the full 16MP. It's just that some of the pixels aren't used.

With M and S modes, there is downsizing occurring and there will some loss in image quality.

Dave

Does this have any effect on a raw file or only on the jpeg? It would be quite a simple thing to simply process the jpeg from a limited part of the raw data.

[rpcrowe]

For me, full size brings slow processing, even on a 4-cpu Core i5 machine, and a image size 4704x3136px that I will never use because I don't print.

That is a reasonable answer and is certainly a valid choice. The smaller image size is certainly enough for any Internet or email use.

The size of my RAW files from my Canon 7D camera is about 5184 x 3456 pixels at 300 pixels per inch...

I have no problems with slow processing on my i7 machine which I have upgraded with maximum RAM (for this older machine running Windows 7 as per my computer tech) as well as an upgraded video card.

I also use hi-speed CF cards (400x and 800x) that are UDMA capable as well as a UDMA capable card reader. My card reader is a Lexar Professional LRW400 Dual Slot SD & CF model. This reader is capable of a USB 3.0 connection although I use it connected to a USB 2.0 slot on my computer.

I don't know if Windows 7 is capable of supporting USB 3.0 connections or if it is, I wonder if the increased speed would be worth the trouble of installing the 3.0 connection.

I don't print a lot of images but, I like to have the best quality originals possible, JUST IN CASE...

[Manfred M]

Does this have any effect on a raw file or only on the jpeg? It would be quite a simple thing to simply process the jpeg from a limited part of the raw data.

Brian - answer is yes. Impact of size setting affects the jpeg, not the raw. Read my posting (just above yours).

[Manfred M]

I don't know if Windows 7 is capable of supporting USB 3.0 connections or if it is, I wonder if the increased speed would be worth the trouble of installing the 3.0 connection.

Yes it does. I installed a USB 3.0 card on my Windows 7 machine and I would highly recommend it, based on my experience. The read rate is dramatically faster than USB 2.0

[Saorsa]

Yes it does. I installed a USB 3.0 card on my Windows 7 machine and I would highly recommend it, based on my experience. The read rate is dramatically faster than USB 2.0

In dealing with large file sizes the amount of physical memory makes a great deal of difference in processing speed.

[dje]

Does this have any effect on a raw file or only on the jpeg? It would be quite a simple thing to simply process the jpeg from a limited part of the raw data.

Brian I don't have a Fuji HS50 to test but I did a test on my Canon G15 which is also a fixed lens camera. I took two RAW shots, one with AR set to 4:3 and the other with AR set to 3:2.

When I opened the two files in ACR, the pixel dimensions were shown as being different (as per the AR) so I concluded from this that the RAW file is cropped in camera. But then I looked at the two files using both RAWDigger and dcRaw and found to my surprise that the files both had the same pixel dimensions corresponding to an AR of 4:3.

My assumption is that the raw files are not cropped in camera but include a tag to indicate to the raw processor software the desired AR. The embedded preview jpeg has been cropped. This is for the Canon of course, I don't know about the Fuji.

Dave

[xpatUSA]

That is a reasonable answer and is certainly a valid choice. The smaller image size is certainly enough for any Internet or email use.

I don't know if Windows 7 is capable of supporting USB 3.0 connections or if it is, I wonder if the increased speed would be worth the trouble of installing the 3.0 connection?

Transferring files from the camera to the PC is not my speed problem. Plus my latest Sigmas are still USB 2.0 and my oldest is 1.0 unreadable by any computer I've got

The speed problem is in the huge size of Sigma X3F files, well over 50MB and the large amount of processing needed by the Foveon captures even just to get decent color. Also Sigma Photo Pro is non-cumulative, i.e. it re-does the image for every adjustment made and has no 'UNDO' feature. However, it remains the best converter for the job.

I don't print a lot of images but, I like to have the best quality originals possible, JUST IN CASE...

I still don't print at all, but I understand your POV.

[Ken MT]

My Canon 7D mk2 has three raw settings,
Setting. Pixels recorded File size

RAW 20M 24Mb
M RAW. 11M. 19.3Mb
S RAW 5M. 13.3Mb
The raw image is raw data output by the image sensor converted to digital data. It is recorded to the card as is and you can select the quality as RAW, M RAW OR S RAW

Don't know if this answers the original query, or makes it moreover confusing.

[Pica]

Hello,
At this stage of my Thinktank on "Camer Sensor - Impact and Utilization", I wish to say thank you, to everyone who has found it stimulating to explore this topic and contributed positively to the thread I created.

To make these your answers justice, I will now contemplate them in my "photographic chamber".

Please feel free to continue adding to the thread and OP.

As always thank you for your cooperation, from Scandinavia!
-

[Saorsa]

Brian I don't have a Fuji HS50 to test but I did a test on my Canon G15 which is also a fixed lens camera. I took two RAW shots, one with AR set to 4:3 and the other with AR set to 3:2.

When I opened the two files in ACR, the pixel dimensions were shown as being different (as per the AR) so I concluded from this that the RAW file is cropped in camera. But then I looked at the two files using both RAWDigger and dcRaw and found to my surprise that the files both had the same pixel dimensions corresponding to an AR of 4:3.

My assumption is that the raw files are not cropped in camera but include a tag to indicate to the raw processor software the desired AR. The embedded preview jpeg has been cropped. This is for the Canon of course, I don't know about the Fuji.

Dave

That's what I thought. It actually takes work (and cpu cycles) to throw data away when all you have to do is make some of the boundary parameters variable parameters.