RAW vs JPG - dynamic range difference?

[Bussty]

So I can confirm that in Photo Mode and using JPEG's there is no upside to using any of the auto modes in terms of Dynamic Range performance.

First image fully manual exposure



Second image ISO switched to Auto



Third image ISO and Shutter Speed to Auto


Fourth image ISO, Shutter Speed, Aperture all Auto



Doesn't seem to be an increase in HDR improvement on any of these so assuming this HDR benefit that you are seeing in d-log M is limited just to Video Mode.

 

[nickcaptures]

Just an update. I updated the firmware of Mavic 4 Pro. Then I did another flight, took some more shots of direct sun during sunset - RAW manual ISO, JPG manual ISO, JPG auto ISO.

On this shoot, the gradient of the light/colour of the RAW shots of the sun looked a lot more natural. I don't know if it was the firmware update, a different position/angle of the camera to the sun, different light intensity or polarisation... but the RAW this time looks far more like I'm used to seeing in other cameras.

As for the JPGs, the dynamic range of the JPGs (using both auto ISO and manual ISO) look the same to me - as what @Bussty found.

Thanks for all the replies everyone.

 

[Meta4]

I don't know if it was the firmware update ...

It's most unlikely that the firmware update made any changes to the camera's properties.
If it did, DJI would have trumpeted that.
It's very rare that firmware updates make any differences to anything that users would notice.

 

[Tolga20]

Your camera applies Local Tone Mapping (LTM) and other image processing algorithms to enhance the live view and JPEG previews (e.g., boosting shadows, balancing highlights, and optimizing overall brightness). However, when you import RAW files into Lightroom or Photoshop, these in-camera enhancements are intentionally not applied, as RAW files are designed to preserve all the original sensor data for maximum editing flexibility.

This is why the shadows may appear darker and highlights more pronounced in Lightroom—you're seeing the unprocessed "raw" data. While this might require some extra editing, it gives you far more control over the final result. Use Auto ISO to gain little bit more dynamic range.

 

[Filmarik]

Your camera applies Local Tone Mapping (LTM) and other image processing algorithms to enhance the live view and JPEG previews (e.g., boosting shadows, balancing highlights, and optimizing overall brightness). However, when you import RAW files into Lightroom or Photoshop, these in-camera enhancements are intentionally not applied, as RAW files are designed to preserve all the original sensor data for maximum editing flexibility.

This is why the shadows may appear darker and highlights more pronounced in Lightroom—you're seeing the unprocessed "raw" data. While this might require some extra editing, it gives you far more control over the final result. Use Auto ISO to gain little bit more dynamic range.

On top of that JPEG has 8bit color depth while DNG has 12 bit color depth. That is a massive difference. That is why DNG allow for much greater flexibility when editing. Auto ISO has nothing to do with dynamic range when shooting DNG.

 

[Greatdane007]

To the OP, firstly is your editing software up to date? Oddly my CaptureOne would only open some of my M4Pro raw files until I updated to the latest version. As for files, a 16 bit raw has about 65,000 pieces of information to work with (or colors). A JPEG file has 256 pieces (or colors). Think of a jpeg as a chocolate cake, and then think of a raw file as; what kind of cake do you want to make. You can not adjust a jpeg beyond the 256 colors within it. In order to gain the most range from a raw file, you would adjust all aspects such as highlights, contrast, shadows, exposure, brightness etc to pull that data to its limits. If you can see colors in a jpeg file, that data and much much more is within the raw data. I think it’s strange that your jpegs “appear better” or have more dynamic range than the raw files. That’s what makes me ask if your software is up to date.

 

[Meta4]

As for files, a 16 bit raw has about 65,000 pieces of information to work with (or colors). A JPEG file has 256 pieces (or colors).

You should check those numbers .. they are a long way from the actual values.

 

[Filmarik]

You should check those numbers .. they are a long way from the actual values.

I guess what he meant to say was that 16bit raw file has approximatelly 65 000 (65,536 to be exact) steps of color per channel whereas 8 bit file such as your typical JPEG has only 256 steps per channel. That would be more or less correct.

 

[GadgetGuy]

I guess what he meant to say was that 16bit raw file has approximatelly 65 000 (65,536 to be exact) steps of color per channel whereas 8 bit file such as your typical JPEG has only 256 steps per channel. That would be more or less correct.

But, didn't you just say DNG only has 12 bit color depth in post #25 above?

 

[Filmarik]

But, didn't you just say DNG only has 12 bit color depth in post #25 above?

Yes. I did.

But, didn't you just say DNG only has 12 bit color depth in post #25 above?

Yes, I did. Again I was talking on behalf of the person who said that a quote: "16 bit RAW file has 65000 information to work with" and I said that that generic statement was close to being accurate. There are not many cameras which shoot 16bit RAW files though. Some MF cameras such as the Fuji GFX 100, as an example does.
AFAIK M4P main camera shoots 14bit RAW files and that means 16,384 steps per channel and M3P shoots RAW files with 12bit color depth meaning 4,096 steps per channel, which is what I was talking about in post #25.
But this debate is about DNG vs. JPEG and all of the above explains and demonstrates why DNG should deliver superior colors to JPEG and more importantly why JPEG breaks appart pretty quickly when pushed in post, and why JPEG is prone to show posterization in smooth gradient areas such as blue sky etc. I can not see any reason why a properly processed DNG file should have inferior colors or less dynamic range than a JPEG.

 

[GadgetGuy]

Yes. I did.

Yes, I did. Again I was talking on behalf of the person who said that a quote: "16 bit RAW file has 65000 information to work with" and I said that that generic statement was close to being accurate. There are not many cameras which shoot 16bit RAW files though. Some MF cameras such as the Fuji GFX 100, as an example does.
AFAIK M4P main camera shoots 14bit RAW files and that means 16,384 steps per channel and M3P shoots RAW files with 12bit color depth meaning 4,096 steps per channel, which is what I was talking about in post #25.
But this debate is about DNG vs. JPEG and all of the above explains and demonstrates why DNG should deliver superior colors to JPEG and more importantly why JPEG breaks appart pretty quickly when pushed in post, and why JPEG is prone to show posterization in smooth gradient areas such as blue sky etc. I can not see any reason why a properly processed DNG file should have inferior colors or less dynamic range than a JPEG.

Thanks for the clarification.

To be clear, then, the only relevant number for the 8 bit JPEG comparison of the Mavic 4 Pro should be 14 bit DNG files, not 16 bit RAW nor 12 bit DNG.

"M4P main camera shoots 14bit RAW files, and that means 16,384 steps per channel instead of 256 steps, assuming DJI's Mavic 4 Pro JPEG's are 8 bit."

Are the Mavic 4 Pro JPEG, in fact, 8 bit?

The poster you were defending was using an irrelevant 16 bit RAW comparison in post #26, and you used an irrelevant 12 bit DNG comparison in post #25.

I am certainly not questioning your premises that DNG should deliver superior color depth, and that JPEG will fall apart rapidly when pushed too far in post, and is prone to show posterization in smooth gradient areas under those circumstances.

The only reason a properly processed DNG might have inferior colors or less dynamic range than a JPEG would be because of some secret sauce of DJI's being applied to the JPEG, that is not available to be used on the DNG's, and/or the DNG was missing the "Colonel’s Secret Sauce."

One of the biggest complaints about Lightroom originally was the DNG's lacked the superior look of the associated camera created JPEG's as a starting point, and Adobe's default opening settings sucked. Eventually, Adobe did their best to copy each camera manufacturers' JPEG processing as a default opening setting, but there are some who still swear each camera manufacturer’s camera JPEG's are superior to what third party image editors can produce from the DNG's when they lack the proprietary "secret sauce" used in the JPEGs.

Bottom line, a well-exposed JPG doesn't need the recovery capability of a DNG. However a DNG gives more flexibility to change the image's color balance and recover areas if an artificial HDR image is desired instead.

There is no right answer as to whether a JPEG or DNG is "better."
It all depends upon what the photographer wants and is trying to achieve, and the compromises required to shoot DNG in the first place (think frame rate) , and whether the editing time and effort required to properly process the DNG after is worth the effort vs. just culling through the JPEG AEB exposures to find the right one, or which to blend together.

 

[Filmarik]

Thanks for the clarification.

To be clear, then, the only relevant number for the 8 bit JPEG comparison of the Mavic 4 Pro should be 14 bit DNG files, not 16 bit RAW nor 12 bit DNG.

"M4P main camera shoots 14bit RAW files, and that means 16,384 steps per channel instead of 256 steps, assuming DJI's Mavic 4 Pro JPEG's are 8 bit."

Are the Mavic 4 Pro JPEG, in fact, 8 bit?

The poster you were defending was using an irrelevant 16 bit RAW comparison in post #26, and you used an irrelevant 12 bit DNG comparison in post #25.

I am certainly not questioning your premises that DNG should deliver superior color depth, and that JPEG will fall apart rapidly when pushed too far in post, and is prone to show posterization in smooth gradient areas under those circumstances.

The only reason a properly processed DNG might have inferior colors or less dynamic range than a JPEG would be because of some secret sauce of DJI's being applied to the JPEG, that is not available to be used on the DNG's, and/or the DNG was missing the "Colonel’s Secret Sauce."

One of the biggest complaints about Lightroom originally was the DNG's lacked the superior look of the associated camera created JPEG's as a starting point, and Adobe's default opening settings sucked. Eventually, Adobe did their best to copy each camera manufacturers' JPEG processing as a default opening setting, but there are some who still swear each camera manufacturer’s camera JPEG's are superior to what third party image editors can produce from the DNG's when they lack the proprietary "secret sauce" used in the JPEGs.

Bottom line, a well-exposed JPG doesn't need the recovery capability of a DNG. However a DNG gives more flexibility to change the image's color balance and recover areas if an artificial HDR image is desired instead.

There is no right answer as to whether a JPEG or DNG is "better."
It all depends upon what the photographer wants and is trying to achieve, and the compromises required to shoot DNG in the first place (think frame rate) , and whether the editing time and effort required to properly process the DNG after is worth the effort vs. just culling through the JPEG AEB exposures to find the right one, or which to blend together.

My only disagreement with you is the last part of your post.
"It all depends upon what the photographer wants and is trying to achieve, and the compromises required to shoot DNG in the first place (think frame rate) , and whether the editing time and effort required to properly process the DNG after is worth the effort vs. just culling through the JPEG AEB exposures to find the right one, or which to blend together."
It is not that difficult to create a dedicated preset in Lr for each camera and then apply that to bunch of images. Timewise not really more time consuming than going through pletora of JPEGs including AEB and culling. As for the frame rate of DNG vs. JPEG this would be a relevant limitation if you were to shoot some fast action like a bird in flight catching a prey or a fast sport action where you would need to shoot for example a high jumper at 30fps and then choose the best frame catching the "decisive moment". But for the vast majority of drone shots? As for AEB and than blending in post that would require about as much time and dedication as processing one DNG, dynamic range of which would be close if not better to about 3 AEB JPEGs and without ghosting issues from moving subjects in the 3 or more AEB frames ( in that hypothetical fast action scenario where framerate would come to question) .
At the end of the day it is a decision of the photographer if he/she wants to shoot JPEG or DNG. This debate started with question about dynamic range of DNG vs. JPEG and the answer to that, from my experience at least is that DNG should be a clear winner in that race.

 

[landscapetog]

One thing to keep in mind is most new TVs and phones/tablets and more and more laptops are now capable of amazing and beautiful HDR image display. It's really stunning and makes an absolutely enormous difference in how images look on any of these devices with more dynamic range (darker blacks, brighter colors and whites).

Photographing with RAW will give you far better results with HDR than trying to do the same thing with JPEGs.

This website will give you examples why you might want to future-proof your images for HDR presentation (assuming you have an HDR display to view these images):

Create and edit true HDR (High Dynamic Range) images - Greg Benz Photography

B&H is currently offering an outstanding deal: $500 off the ASUS PA32UCXR 32" 4K HDR monitor, which is a stunning 1600 nits HDR monitor with 2304 local dimming zones (4x the Pro Display XDR), automated calibration, and single cable charging/pass-through. See my full reviews and other HDR monitor...

gregbenzphotography.com

 

[Greatdane007]

I took a Capture One course some years ago in which we learned about RAW files vs JPEGs. Long story short, a raw file contains up to 60,000 pieces of data or colors. There are only 256 colors max in a JPEG file. The RAW file colors can be tweaked and toned in an unlimited direction of colors, brightness, contrast, etc, where the JPEG is pretty much set in stone without much wiggle room. An HDR blended image would bring in some added options that might not be there frame to frame but you are still maxed at 256 colors with a Josh. I only shoot RAW, process and edit, and then output to TIFF (4056 colors) or JPEG, depending upon image applications. A print magazine can produce far more colors than your phone screen.

 

[anotherlab]

I took a Capture One course some years ago in which we learned about RAW files vs JPEGs. Long story short, a raw file contains up to 60,000 pieces of data or colors. There are only 256 colors max in a JPEG file. The RAW file colors can be tweaked and toned in an unlimited direction of colors, brightness, contrast, etc, where the JPEG is pretty much set in stone without much wiggle room. An HDR blended image would bring in some added options that might not be there frame to frame but you are still maxed at 256 colors with a Josh. I only shoot RAW, process and edit, and then output to TIFF (4056 colors) or JPEG, depending upon image applications. A print magazine can produce far more colors than your phone screen.

JPEG supports 24-bit colors or 16.7 million distinct color combinations. Each color channel (red, green, blue ) supports 8 bit values for intensities. That combination yields 16.7 million combinations.

The color support for RAW will depend on the bit depth per channel. RAW isn't a single format; it's a label used to describe vendor-specific implementations. A RAW file will typically support between 68 billion to 4 trillion distinct colors. Not that any of us can distinguish that many distinct colors.

TIFF has support for different color depths. From 8-bit gray to 24-bit, 32-bit, or even 42-bit color.

FWIW. an iPhone screen can display more colors than a print magazine. The current crop of iPhones uses the Display P3 color space, which defines more colors than the CYMK space used for print. The iPhone screens are also calibrated during assembly,

 

[Meta4]

There are only 256 colors max in a JPEG file. ... I only shoot RAW, process and edit, and then output to TIFF (4056 colors)

Neither of those numbers are anywhere near the truth.

 

[kanesimages]

Hi everyone,

New owner of a Mavic 4 Pro here. I was testing capturing photos directly into the sun, and for some reason, I'm finding that the JPG photo is giving a much more pleasing representation of direct sun glare than the RAW photo. Any ideas why this would be?

I am already aware that shooting video in D-LOG M is giving superior dynamic range to D-LOG (as this video explains:

To view this content we will need your consent to set third party cookies.
For more detailed information, see our cookies page.

Accept third party cookies

)

Could the same software / dual ISO trickery be also happening with the Mavic 4 photo mode to make it that shooting JPG is superior to shooting RAW in high dynamic range situations... (like direct sun?). Or am I doing something wrong with my RAW photos?

Any ideas/suggestions would be welcome. I'm baffled.

I'll attach examples here.

The original DNG file is 127MG & the original JPG file is 47MB. So I've put them both into photoshop, reduced by 80%, and exported as jpgs. While this isn't the original files, it's still a representation of the difference. Look at the sun and the range of colour/light outwards from the sun's centre in both images. The DNG seems to have like 3 colours/intensity of light. The JPG has a much more pleasant gradient of colour and light. This is more than just increased saturation...

Any idea what is going on? Thanks so much.

I haven’t taken the time read all these lengthy replies but the JPEG appears to have been hit up with some fairly significant NR and overall warmth. Maybe the drone is blending a NR file with the Raw. A lot of processors go to NR in post processing if they want a less subtle variation than using a Gaussian blur.