jpg question - compression / quality settings comparing different size sensors
- Thread starter Texas
- Start date
Not sure of your definition of "same file size", or if it is about same compression properties, like both Large or Medium.
RGB images (speaking of JPG) are 3 bytes per pixel (24-bit color, RGB), so any 24 mp image actual data size is 72 MB (roughly) and any 12 mp image is 36 MB (roughly, because 1 MB is actually 1.049 million bytes).
That is the actual data size (when opened in computer memory), but then JPG compression drastically reduces the size while in the file.
D7200, 24 mp. Manual says Large JPG is approx 12.7 MB, and Medium is 7.2 MB.
D300, 12 mp. Manual says Large JPG is approx 5.8 MB, and Medium is 3.3 MB.
These sizes will vary with image content. Very highly detailed images (image of a tree full of leaves) will be larger, but bland featureless images (sky or a blank wall) will compress much smaller. A folder of many varied "same size" images will often show about a 2:1 size range of the files.
But note these example numbers are proportional. With same compression, then 24 mp is about twice the file size of 12 mp. Because the number of pixels is 2x.
So, if your "same file size" means actual file size in bytes, then of course my bet is on the 24 mp image, with twice the pixels and half of the JPG compression.
But two Large images of same scene should look comparable. Depends really on how it is viewed. Viewed on our computer screen, both are normally resampled much smaller (2 mp will fill a 1920x1080 screen) to fit them onto the screen. 24 mp won't have the advantage then that it has when viewed at 100% Actual size.
Last edited:
The Nikon manuals give a chart of nominal file size given the three resolution and three sizes Large....Basic, etc.
Yes I'm talking nominal file size vs. IQ - another way to ask the question would be "is there any detectable difference, given same size prints, from a high pixel count sensor that has been used to make a small basic jpg vs. a camera with a low pixel count sensor running at Large / Fine" where the jpg file sizes are about same.
Or even simpler, do I need a D5 to make the very best jpgs for the web ?
I think I know the answer but never seen such a test run by those who write photo articles.
Yes I'm talking nominal file size vs. IQ - another way to ask the question would be "is there any detectable difference, given same size prints, from a high pixel count sensor that has been used to make a small basic jpg vs. a camera with a low pixel count sensor running at Large / Fine" where the jpg file sizes are about same.
Or even simpler, do I need a D5 to make the very best jpgs for the web ?
I think I know the answer but never seen such a test run by those who write photo articles.
Well, Large, Medium and Small are image sizes, different numbers of pixels.
Fine, Normal, and Basic are JPG Quality parameters, which does not change pixel count, but which offers a lower image quality.
Both can reduce file size, in different ways. Image quality and resolution are not the same properties.
So Large and Basic are not on the same concept scale.
Camera default is Large Fine.
If concerned with better image quality, there would seem to be no question about it, select Fine. If concerned about quality, why would you want less quality? Is file size that important? Disks and cards are inexpensive compared to the camera, or even lenses. Would you want to instead see your alternative result choices before forcing that choice?
If concerned with image size, then 24 mp sizes are:
Large 6000x4000
Medium 4496x3000
Small 2992x2000
Now if you are always going to resample even smaller to view on a 1920x1080 monitor screen, or if going to print a 6x4 inch print at 1800x1200 pixels, then I frankly doubt it will make much difference which size you start from. However, the larger size can offer larger prints with greater resolution, and also offers greatly more choice about cropping some of it away, and still having sufficient pixels for your purpose.
And we don't always realize when our next picture will be the keeper of a lifetime...
I would suggest always Large Fine (because there is no going back later), and then when and if you must have a smaller file, simply handle it later. Maybe save and archive the original first.
Last edited:
Bikerbrent
Senior Member
And we don't always realize when our next picture will be the keeper of a lifetime...
I would suggest always Large Fine (because there is no going back later), and then when and if you must have a smaller file, simply handle it later. Maybe save and archive the original first.
I concur strongly with Wayne's comments. I jumped from a D200 to a D7200 and have never once considered reducing quality of the D7200 images. In fact, I shoot almost exclusively in raw. I would only consider JPEG if I was shooting some action sport that required a higher frame buffer, and that is almost never.
Yes I understand how jpeg settings work and what they do and how, as well as Nikon's 'optimize file size' setting, file size vs. subject detail and and have seen Ken R's comparisons of IQ settings using different jpeg settings, but those are all on the same camera.
Simplifying further:
I'm asking about the relative IQ of jpegs from 2 different cameras, one with a bigger sensor set for higher compression than the one from a smaller sensor set for Large/Fine. Size and compression being set to give pretty close to the same file sizes from both cameras.
How do the two similar file size, but different compression jpegs compare ?
The two being a more highly compressed jpg from a 12, 16, 24, or greater mb sensor vs. the 'fine' (most lightly) compressed jpeg from a 6mb sensor - choosing Q factors on the large sensor camera (size and res) and lightest compression (meaning fine) on the 6mb such that the final jpeg file sizes are comparable.
Theory is fine, and may be all that's out there relegating this to a thought experiment.
But with all the interesting things folks have reported on in this forum I figured someone has already done this experiment.
I'd like to think about two different cameras that differ only in sensor size but that is not likely possible.
No doubt there are other factors involved between two different cameras that could skew this jpeg IQ comparison experiment:
1/ the actual jpeg generation math engine/demosaic'er
2/ pixel quality/size vs overall pixel count
3/ lots of others
Simplifying further:
I'm asking about the relative IQ of jpegs from 2 different cameras, one with a bigger sensor set for higher compression than the one from a smaller sensor set for Large/Fine. Size and compression being set to give pretty close to the same file sizes from both cameras.
How do the two similar file size, but different compression jpegs compare ?
The two being a more highly compressed jpg from a 12, 16, 24, or greater mb sensor vs. the 'fine' (most lightly) compressed jpeg from a 6mb sensor - choosing Q factors on the large sensor camera (size and res) and lightest compression (meaning fine) on the 6mb such that the final jpeg file sizes are comparable.
Theory is fine, and may be all that's out there relegating this to a thought experiment.
But with all the interesting things folks have reported on in this forum I figured someone has already done this experiment.
I'd like to think about two different cameras that differ only in sensor size but that is not likely possible.
No doubt there are other factors involved between two different cameras that could skew this jpeg IQ comparison experiment:
1/ the actual jpeg generation math engine/demosaic'er
2/ pixel quality/size vs overall pixel count
3/ lots of others
Bikerbrent
Senior Member
How about several years to technical development of the sensor itself? They are NOT going to differ only in sensor size. I'm done here, life is too short for low IQ.
In camera jpeg are generated by camera software, and each manufacturer has his own recipe, so do not expect jpegs of te same scene to be identical across cameras.
Jpeg files are always lossy, hence they loose quality right at the source, as well as every time you edit and save. As mentioned in previous posts, disk and memory is cheap so why compromise? Even if you have to down sample from 24mp to 4mp, the down sampled image will in generally be better than the image saved at 4mp.
If you are concerned about quality, then shoot RAW, and use Nikon View NXi to batch convert to jpeg. In case you want to process a few images further (recover shadows, adjust contrast or exposure) RAW will give much better results.
Jpeg files are always lossy, hence they loose quality right at the source, as well as every time you edit and save. As mentioned in previous posts, disk and memory is cheap so why compromise? Even if you have to down sample from 24mp to 4mp, the down sampled image will in generally be better than the image saved at 4mp.
If you are concerned about quality, then shoot RAW, and use Nikon View NXi to batch convert to jpeg. In case you want to process a few images further (recover shadows, adjust contrast or exposure) RAW will give much better results.
Last edited:
RocketCowboy
Senior Member
IMO, most of all this comes down to the starting image, before JPeg processing is applied. Lower starting IQ will result in lower finishing IQ.
The missing piece, the jpeg processing portion of the equation, is also improving as sensor enhancements come down the road. The image processor in that 12MP camera doesn't crank out the same quality output as the image processor in the 24MP camera. While the starting RAW image has gotten better, so has the processor and processing within the camera. Downsampling the 24MP image to be a 12MP image, and then processing it into JPG ... I would expect the 24MP body to still provide a superior image, simply because there was more data to work with from the start.
The missing piece, the jpeg processing portion of the equation, is also improving as sensor enhancements come down the road. The image processor in that 12MP camera doesn't crank out the same quality output as the image processor in the 24MP camera. While the starting RAW image has gotten better, so has the processor and processing within the camera. Downsampling the 24MP image to be a 12MP image, and then processing it into JPG ... I would expect the 24MP body to still provide a superior image, simply because there was more data to work with from the start.
RocketCowboy
Senior Member
Human eyes are better than our 'digital' eyes, otherwise we wouldn't need things like white balance, HDR, etc. The challenge with your question is that both sides of the quality comparison are riding technology trends. Pixel density increases as the technology evolves, and likewise the image processing capabilities (i.e. jpeg encoders) also improve as the technology evolves.
Don Kuykendall_RIP
RIP :(
No matter what the camera use the best settings (Fine/large) if you are limiting yourself to JPEG. You pay big bucks for a better camera why would you want to limit its performance. That is why I only shoot RAW and process every photo I use so I know I have the best results I can get.
Horoscope Fish
Senior Member
I don't think the question can be answered; not as you've posed it.
Here's an answer to the question, and as a plus, it removes the need for two cameras with all their differences beyond sensor size. Answers the compression / resolution trade off nicely:
Higher resolution or lower compression JPG?
Higher resolution or lower compression JPG?
Fat Fingers
Banned
For the OP. consider this ....just shot 40 weddings this year all with D7100/810 all jpeg basic sharp +9 f8 auto iso ...
Never had a problem or complaint . Why mess about just get on with it ....you cannot process 1500 RAW pics individually from a wedding you have to baulk convert so why not let the software in the camera do it ...
Never had a problem or complaint . Why mess about just get on with it ....you cannot process 1500 RAW pics individually from a wedding you have to baulk convert so why not let the software in the camera do it ...
I don't know why folks had such a hard time with my initial question, re-cast and further simplified twice more.
Anyway my curiosity was about the relative IQ's of a hi res jpg compressed 2 or 4x more than a lower res / but less compressed jpg in order to yield similar file sizes. A little googling answered the question perfectly yielding the link I posted a few msgs back.
No need to tell me how set my camera to make pictures. That's a subject for another thread.
Anyway my curiosity was about the relative IQ's of a hi res jpg compressed 2 or 4x more than a lower res / but less compressed jpg in order to yield similar file sizes. A little googling answered the question perfectly yielding the link I posted a few msgs back.
No need to tell me how set my camera to make pictures. That's a subject for another thread.
Horoscope Fish
Senior Member
That's a wedding every eight days.You shot 40 weddings at f/8? Interesting.