RAW, HEIF, and JPEG: TN Strikes Again

[Thomas A Anderson]

Some more dubious analysis from Tony Northrup. I don’t recommend watching the video because his analysis is so utterly skewed and useless, but at this point he’s got so many subscribers I don’t think avoiding him as a subject will do any good.

Tony sings the praises of HEIF/HEIC as a replacement for JPEG and then supports his conclusion with poorly reasoned or intentionally misleading nonsense scare tactics. He discusses the cost/benefit ratio in terms of JPEG or RAW files collected over a lifetime of shooting……and then assumes that total number as a cost to back up from day one. Assuming a linear progression, calculating a cloud backup cost that slowly increases in capacity over years or decades wouldn’t be hard to do. Heck, you could even do that analysis on a month-by-month basis with some fairly rudimentary formulas in Excel using a brute force approach.

So, he starts by talking about shooting 100,000 images over a career (conservative, so that’s something at least) and then saying the best way to back them up is on the cloud. That may be convenient, but anyone who wants to control their images and avoid general release to the entire world via the inevitable hack will just buy a few hard drives and institute a fairly straight forward backup plan. And hard drives get cheaper and better as time goes on, so the value proposition only improves over time.

He then goes on to say that using a lossless compression system for RAW files is basically useless and stupid. You’ll note that at the beginning he says this isn’t going to be yet another RAW vs. JPEG video. That lasted about seven minutes. Some more words come out of his face once again saying how dumb RAW is and how great compressed files (JPEG and now HEIF) are.

And the icing on the cake is going to the camera storage and using the most expensive SD cards available to compare costs. He then goes on to say the buffer will get extended by 2X with a file that is half the size, ignoring that only the cheapest cameras have any real JPEG buffer wall.

Honestly, Tony is moving towards Ken Rockwell’s side of the spectrum in terms of the extremely poor logic used to justify poor conclusions.

 

[ZX11]

The entertainment industry's theory is that anything can be successfully argued. Especially, in light of the modern argument techniques in use.

The more the argument flies in the face of conventional wisdom, the more entertaining it is. The argument no longer has to be well supported to be published/successful. Hence, RAW can be said to suck and the earth is flat.

The Apple iPhone is the best camera,...ever!

 

[NoTicket]

I would set aside Tony’s views from the JPEG -> HEIF or AVIF transition.

JPEG is a truly limiting medium now that displays are reaching higher capabilities.

We have finally reached the point in which wide gamut color spaces are capable of being displayed by an increasing number of screens.

Using Display P3, you get a 25% wider color gamut, which also means to get the same precision that you got with an 8bit sRGB image, you’re going to need a 9th bit or face increased banding.

Using the BT.2020 color space the problem gets even worse as it is even wider.

The reason we need HEIF or AVIF is because it allows still image capture and transmission to be up to par with the color resolution of video, instead of being stuck in the past.

I still disagree with Tony Northrup on the concept that one can swap out RAW for HEIF. HEIF compression is just as bad and as lossy as JPEG. Either format loses the scene reference data as the white balancing and color processing is applied through the pipeline.

 

[ZX11]

I would set aside Tony’s views from the JPEG -> HEIF or AVIF transition.

JPEG is a truly limiting medium now that displays are reaching higher capabilities.

We have finally reached the point in which wide gamut color spaces are capable of being displayed by an increasing number of screens.

Using Display P3, you get a 25% wider color gamut, which also means to get the same precision that you got with an 8bit sRGB image, you’re going to need a 9th bit or face increased banding.

Using the BT.2020 color space the problem gets even worse as it is even wider.

The reason we need HEIF or AVIF is because it allows still image capture and transmission to be up to par with the color resolution of video, instead of being stuck in the past.

I still disagree with Tony Northrup on the concept that one can swap out RAW for HEIF. HEIF compression is just as bad and as lossy as JPEG. Either format loses the scene reference data as the white balancing and color processing is applied through the pipeline.

That is interesting what new display tech is driving with photo files.

When you send someone a picture to view, do you send JPEG, RAW, or something else? All my gallery pics are JPEG. And, I still run into the problem that my photos are viewed by friends and family on old smartphones.

 

[NoTicket]

I would set aside Tony’s views from the JPEG -> HEIF or AVIF transition.

JPEG is a truly limiting medium now that displays are reaching higher capabilities.

We have finally reached the point in which wide gamut color spaces are capable of being displayed by an increasing number of screens.

Using Display P3, you get a 25% wider color gamut, which also means to get the same precision that you got with an 8bit sRGB image, you’re going to need a 9th bit or face increased banding.

Using the BT.2020 color space the problem gets even worse as it is even wider.

The reason we need HEIF or AVIF is because it allows still image capture and transmission to be up to par with the color resolution of video, instead of being stuck in the past.

I still disagree with Tony Northrup on the concept that one can swap out RAW for HEIF. HEIF compression is just as bad and as lossy as JPEG. Either format loses the scene reference data as the white balancing and color processing is applied through the pipeline.

That is interesting what new display tech is driving with photo files.

When you send someone a picture to view, do you send JPEG, RAW, or something else? All my gallery pics are JPEG. And, I still run into the problem that my photos are viewed by friends and family on old smartphones.

 

[J A C S]

Honestly, Tony is moving towards Ken Rockwell’s side of the spectrum in terms of the extremely poor logic used to justify poor conclusions.

Don't insult KR. He is entertaining at least and knows classical gear well.

 

[shawnphoto]

Some more dubious analysis from Tony Northrup. I don’t recommend watching the video because his analysis is so utterly skewed and useless, but at this point he’s got so many subscribers I don’t think avoiding him as a subject will do any good.

Tony sings the praises of HEIF/HEIC as a replacement for JPEG and then supports his conclusion with poorly reasoned or intentionally misleading nonsense scare tactics. He discusses the cost/benefit ratio in terms of JPEG or RAW files collected over a lifetime of shooting……and then assumes that total number as a cost to back up from day one. Assuming a linear progression, calculating a cloud backup cost that slowly increases in capacity over years or decades wouldn’t be hard to do. Heck, you could even do that analysis on a month-by-month basis with some fairly rudimentary formulas in Excel using a brute force approach.

So, he starts by talking about shooting 100,000 images over a career (conservative, so that’s something at least) and then saying the best way to back them up is on the cloud. That may be convenient, but anyone who wants to control their images and avoid general release to the entire world via the inevitable hack will just buy a few hard drives and institute a fairly straight forward backup plan. And hard drives get cheaper and better as time goes on, so the value proposition only improves over time.

He then goes on to say that using a lossless compression system for RAW files is basically useless and stupid. You’ll note that at the beginning he says this isn’t going to be yet another RAW vs. JPEG video. That lasted about seven minutes. Some more words come out of his face once again saying how dumb RAW is and how great compressed files (JPEG and now HEIF) are.

And the icing on the cake is going to the camera storage and using the most expensive SD cards available to compare costs. He then goes on to say the buffer will get extended by 2X with a file that is half the size, ignoring that only the cheapest cameras have any real JPEG buffer wall.

Honestly, Tony is moving towards Ken Rockwell’s side of the spectrum in terms of the extremely poor logic used to justify poor conclusions.

Think about if your flash fails to fire or you just simply make a mistake. Having the RAW file gives you the option of recovering a bad exposure to a good state.

If you think you can get every shot *perfect* or you just don't care how your photos look then by all means shoot .jpg.

 

[Thomas A Anderson]

The entertainment industry's theory is that anything can be successfully argued. Especially, in light of the modern argument techniques in use.

The more the argument flies in the face of conventional wisdom, the more entertaining it is. The argument no longer has to be well supported to be published/successful. Hence, RAW can be said to suck and the earth is flat.

The Apple iPhone is the best camera,...ever!

Is for a decent argument. Tony has influence, for good or bad, but his analysis and support for his arguments are somewhere between thin, non-existent, and outright deceptive. I’m not sure if he’s intentionally deceiving or simply wrong but he still leads to these bizarre, wrong headed conclusions.

I posted a long list of jpeg and raw myths in the M forum a while back motivated by Tony and people in that forum using horrible arguments for and against both. Funny that the person who claimed everything I said was wrong never told me what I was wrong about.

 

[ZX11]

The entertainment industry's theory is that anything can be successfully argued. Especially, in light of the modern argument techniques in use.

The more the argument flies in the face of conventional wisdom, the more entertaining it is. The argument no longer has to be well supported to be published/successful. Hence, RAW can be said to suck and the earth is flat.

The Apple iPhone is the best camera,...ever!

Is for a decent argument. Tony has influence, for good or bad, but his analysis and support for his arguments are somewhere between thin, non-existent, and outright deceptive. I’m not sure if he’s intentionally deceiving or simply wrong but he still leads to these bizarre, wrong headed conclusions.

I posted a long list of jpeg and raw myths in the M forum a while back motivated by Tony and people in that forum using horrible arguments for and against both. Funny that the person who claimed everything I said was wrong never told me what I was wrong about.

I think he needs to push a controversial topic in order to get views/attention. But that leaves him trying to support a point that is mostly unsupportable.

I should do a video on why IBIS is bad. Support #1: I have other, and better, ways of stabilizing the camera so I don't need a sensor floating around. Support #2: ........

 

[Graham C F]

There are two separate questions here:

1. Should we shoot in RAW, or in a compressed format?

For people wanting the best possible quality and editing latitude, the answer is clearly RAW or C-RAW. For people, such as press photographers or social media posters, needing speed, JPEG is the answer.

2. What format should we use for sharing and displaying photos?

In this case, we need a universal standard - ideally one that is manufacturer independent, or at least supported by all of them. JPEG obviously achieves this, but it sacrifices some of the dynamic and colour range that modern cameras and displays are capable of. Sooner or later, customers will demand something better. Therefore, I would say it is inevitable that we will migrate to a more advanced compression format. Whether that is HEIF/HEIC or something else, I am not qualified to judge.

 

[Leon Wittwer]

I'm not sure that color range is hindered by jpeg. It seems to me that the spacing between color values becomes very slightly larger as the color gamut becomes larger. If the gradations are still smaller than what the eye can see, then an image should display just fine. The eye is likely the limiting factor. Jpeg likely be the defect standard for some time to come.

 

[Graham C F]

I'm not sure that color range is hindered by jpeg. It seems to me that the spacing between color values becomes very slightly larger as the color gamut becomes larger. If the gradations are still smaller than what the eye can see, then an image should display just fine. The eye is likely the limiting factor. Jpeg likely be the defect standard for some time to come.

Hi Leon,

Having done a little investigation, I have discovered that this is a topic that could (and presumably has) fill many PhD theses!

I have found an interesting dpreview forum discussion here:

https://www.dpreview.com/forums/thread/3386276#forum-post-50880314

and some interesting explanations here:

https://petapixel.com/2018/12/13/srgb-vs-adobe-rgb-vs-prophoto-rgb-color-spaces-explained/

Graham

 

[Peak freak]

I shoot RAW for all the 'usual' reasons, but one of the most useful and indispensable, is the ability to adjust colour temperature. Do these new compression formats support post WB adjustment? If not, let the likes of TN miss out on such a powerful and creative option.

[Edit: Another thing I have noticed (but more specialized) is that 8bit files don't stitch that well in PS with posterization artifacts - enough to see when printed.]

 

[Thomas A Anderson]

I shoot RAW for all the 'usual' reasons, but one of the most useful and indispensable, is the ability to adjust colour temperature. Do these new compression formats support post WB adjustment? If not, let the likes of TN miss out on such a powerful and creative option.

[Edit: Another thing I have noticed (but more specialized) is that 8bit files don't stitch that well in PS with posterization artifacts - enough to see when printed.]

The only way you’ll ever be able to adjust white balance without further narrowing your post processing options will be in RAW. 10-bits per channel may be better than 8-bit jpeg but it’s still not 14. And lossy compression will always be lossy. It would be great if HEIF improves on jpeg, but RAW will always be your digital negative.

 

[Peak freak]

I shoot RAW for all the 'usual' reasons, but one of the most useful and indispensable, is the ability to adjust colour temperature. Do these new compression formats support post WB adjustment? If not, let the likes of TN miss out on such a powerful and creative option.

[Edit: Another thing I have noticed (but more specialized) is that 8bit files don't stitch that well in PS with posterization artifacts - enough to see when printed.]

The only way you’ll ever be able to adjust white balance without further narrowing your post processing options will be in RAW. 10-bits per channel may be better than 8-bit jpeg but it’s still not 14. And lossy compression will always be lossy. It would be great if HEIF improves on jpeg, but RAW will always be your digital negative.

Thanks. Not wanting to be picky but... RAW can output at 16 bpc (not 14).

I'm sure you are aware, but those two (or more) bits make a big difference, approx. 65,000 tones vs 16,000 (14bit). So while we are at it, standard 8bit is 256, and 10 bit 1280 tones.

I don't doubt that for a lot of applications, particularly web, 8 or 10 bit will suffice, but I'll take as much processing latitude as I can have. I see it as my opportunity to make something better.

Cheers.

 

[Thomas A Anderson]

I shoot RAW for all the 'usual' reasons, but one of the most useful and indispensable, is the ability to adjust colour temperature. Do these new compression formats support post WB adjustment? If not, let the likes of TN miss out on such a powerful and creative option.

[Edit: Another thing I have noticed (but more specialized) is that 8bit files don't stitch that well in PS with posterization artifacts - enough to see when printed.]

The only way you’ll ever be able to adjust white balance without further narrowing your post processing options will be in RAW. 10-bits per channel may be better than 8-bit jpeg but it’s still not 14. And lossy compression will always be lossy. It would be great if HEIF improves on jpeg, but RAW will always be your digital negative.

Thanks. Not wanting to be picky but... RAW can output at 16 bpc (not 14).

What camera uses 16 bit RAW files? Every Canon I’m aware of max’s out at 14 and some drop to 12 in continuous shooting.

I don’t mind one being picky, but it helps to be right.

I'm sure you are aware, but those two (or more) bits make a big difference, approx. 65,000 tones vs 16,000 (14bit). So while we are at it, standard 8bit is 256, and 10 bit 1280 tones.

I don't doubt that for a lot of applications, particularly web, 8 or 10 bit will suffice, but I'll take as much processing latitude as I can have. I see it as my opportunity to make something better.

Cheers.

 

[Peak freak]

I shoot RAW for all the 'usual' reasons, but one of the most useful and indispensable, is the ability to adjust colour temperature. Do these new compression formats support post WB adjustment? If not, let the likes of TN miss out on such a powerful and creative option.

[Edit: Another thing I have noticed (but more specialized) is that 8bit files don't stitch that well in PS with posterization artifacts - enough to see when printed.]

The only way you’ll ever be able to adjust white balance without further narrowing your post processing options will be in RAW. 10-bits per channel may be better than 8-bit jpeg but it’s still not 14. And lossy compression will always be lossy. It would be great if HEIF improves on jpeg, but RAW will always be your digital negative.

Thanks. Not wanting to be picky but... RAW can output at 16 bpc (not 14).

What camera uses 16 bit RAW files? Every Canon I’m aware of max’s out at 14 and some drop to 12 in continuous shooting.

I don’t mind one being picky, but it helps to be right.

I'm sure you are aware, but those two (or more) bits make a big difference, approx. 65,000 tones vs 16,000 (14bit). So while we are at it, standard 8bit is 256, and 10 bit 1280 tones.

I don't doubt that for a lot of applications, particularly web, 8 or 10 bit will suffice, but I'll take as much processing latitude as I can have. I see it as my opportunity to make something better.

Cheers.

Sorry, I am not talking about the in camera RAW capture (eg 12/14bit).

Rather, I am referring to the post processing flexibility of shooting RAW files.

Shooting in RAW format allows you to process your image in 8 or 16 bit format, that is, with 256 tones (2*8) or 65,536 tones (2*16) for much smoother gradations. Jpeg is limited to 8bit, end of story.

I am not sure about newer formats like HEIC/F, but I guess it is only 8bit as it is derived from video which is usually 8-10bit compression.

As a practical example, once you start editing an 8bit file, you start to see gaps in your histogram where a tone level drops out - leading to posterization. The same edits in 16bit mode contain enough tone gradations to 'fill in the missing gaps' when you update your histogram.

One thing to be aware of is that your editing program will often display lower quality (8bit) previews that show the posterization. Zoom in and it will disappear as the true (16bit) tones are displayed.

Suffice to say I haven't seen Tony's video, but I think we agree that compression is good, and bad, depending on what you want.

 

[NoTicket]

I shoot RAW for all the 'usual' reasons, but one of the most useful and indispensable, is the ability to adjust colour temperature. Do these new compression formats support post WB adjustment? If not, let the likes of TN miss out on such a powerful and creative option.

[Edit: Another thing I have noticed (but more specialized) is that 8bit files don't stitch that well in PS with posterization artifacts - enough to see when printed.]

The only way you’ll ever be able to adjust white balance without further narrowing your post processing options will be in RAW. 10-bits per channel may be better than 8-bit jpeg but it’s still not 14. And lossy compression will always be lossy. It would be great if HEIF improves on jpeg, but RAW will always be your digital negative.

Thanks. Not wanting to be picky but... RAW can output at 16 bpc (not 14).

What camera uses 16 bit RAW files? Every Canon I’m aware of max’s out at 14 and some drop to 12 in continuous shooting.

I don’t mind one being picky, but it helps to be right.

I'm sure you are aware, but those two (or more) bits make a big difference, approx. 65,000 tones vs 16,000 (14bit). So while we are at it, standard 8bit is 256, and 10 bit 1280 tones.

I don't doubt that for a lot of applications, particularly web, 8 or 10 bit will suffice, but I'll take as much processing latitude as I can have. I see it as my opportunity to make something better.

Cheers.

The Fuji GFX 100 Not a Canon camera though.

 

[Thomas A Anderson]

I shoot RAW for all the 'usual' reasons, but one of the most useful and indispensable, is the ability to adjust colour temperature. Do these new compression formats support post WB adjustment? If not, let the likes of TN miss out on such a powerful and creative option.

[Edit: Another thing I have noticed (but more specialized) is that 8bit files don't stitch that well in PS with posterization artifacts - enough to see when printed.]

The only way you’ll ever be able to adjust white balance without further narrowing your post processing options will be in RAW. 10-bits per channel may be better than 8-bit jpeg but it’s still not 14. And lossy compression will always be lossy. It would be great if HEIF improves on jpeg, but RAW will always be your digital negative.

Thanks. Not wanting to be picky but... RAW can output at 16 bpc (not 14).

What camera uses 16 bit RAW files? Every Canon I’m aware of max’s out at 14 and some drop to 12 in continuous shooting.

I don’t mind one being picky, but it helps to be right.

I'm sure you are aware, but those two (or more) bits make a big difference, approx. 65,000 tones vs 16,000 (14bit). So while we are at it, standard 8bit is 256, and 10 bit 1280 tones.

I don't doubt that for a lot of applications, particularly web, 8 or 10 bit will suffice, but I'll take as much processing latitude as I can have. I see it as my opportunity to make something better.

Cheers.

Sorry, I am not talking about the in camera RAW capture (eg 12/14bit).

Rather, I am referring to the post processing flexibility of shooting RAW files.

Post processing uses exactly 14 bits per color channel. That’s all the data you have to work with. The Analog-to-Digital-Conversion is 14 bits, so 2^14 levels per channel is it. 16,384 is your max and 0 is your min no matter what the file package used happens to be capable of. You can save it in a 16 bit TIFF file but there is no physical way to invent the additional data; it would have to be encoded at 16 bit upon capture.

Shooting in RAW format allows you to process your image in 8 or 16 bit format, that is, with 256 tones (2*8) or 65,536 tones (2*16) for much smoother gradations. Jpeg is limited to 8bit, end of story.

I am not sure about newer formats like HEIC/F, but I guess it is only 8bit as it is derived from video which is usually 8-10bit compression.

As a practical example, once you start editing an 8bit file, you start to see gaps in your histogram where a tone level drops out - leading to posterization. The same edits in 16bit mode contain enough tone gradations to 'fill in the missing gaps' when you update your histogram.

One thing to be aware of is that your editing program will often display lower quality (8bit) previews that show the posterization. Zoom in and it will disappear as the true (16bit) tones are displayed.

Suffice to say I haven't seen Tony's video, but I think we agree that compression is good, and bad, depending on what you want.

 

[NoTicket]

I shoot RAW for all the 'usual' reasons, but one of the most useful and indispensable, is the ability to adjust colour temperature. Do these new compression formats support post WB adjustment? If not, let the likes of TN miss out on such a powerful and creative option.

[Edit: Another thing I have noticed (but more specialized) is that 8bit files don't stitch that well in PS with posterization artifacts - enough to see when printed.]

The only way you’ll ever be able to adjust white balance without further narrowing your post processing options will be in RAW. 10-bits per channel may be better than 8-bit jpeg but it’s still not 14. And lossy compression will always be lossy. It would be great if HEIF improves on jpeg, but RAW will always be your digital negative.

Thanks. Not wanting to be picky but... RAW can output at 16 bpc (not 14).

What camera uses 16 bit RAW files? Every Canon I’m aware of max’s out at 14 and some drop to 12 in continuous shooting.

I don’t mind one being picky, but it helps to be right.

I'm sure you are aware, but those two (or more) bits make a big difference, approx. 65,000 tones vs 16,000 (14bit). So while we are at it, standard 8bit is 256, and 10 bit 1280 tones.

I don't doubt that for a lot of applications, particularly web, 8 or 10 bit will suffice, but I'll take as much processing latitude as I can have. I see it as my opportunity to make something better.

Cheers.

Sorry, I am not talking about the in camera RAW capture (eg 12/14bit).

Rather, I am referring to the post processing flexibility of shooting RAW files.

Post processing uses exactly 14 bits per color channel. That’s all the data you have to work with. The Analog-to-Digital-Conversion is 14 bits, so 2^14 levels per channel is it. 16,384 is your max and 0 is your min no matter what the file package used happens to be capable of. You can save it in a 16 bit TIFF file but there is no physical way to invent the additional data; it would have to be encoded at 16 bit upon capture.

Shooting in RAW format allows you to process your image in 8 or 16 bit format, that is, with 256 tones (2*8) or 65,536 tones (2*16) for much smoother gradations. Jpeg is limited to 8bit, end of story.

I am not sure about newer formats like HEIC/F, but I guess it is only 8bit as it is derived from video which is usually 8-10bit compression.

As a practical example, once you start editing an 8bit file, you start to see gaps in your histogram where a tone level drops out - leading to posterization. The same edits in 16bit mode contain enough tone gradations to 'fill in the missing gaps' when you update your histogram.

One thing to be aware of is that your editing program will often display lower quality (8bit) previews that show the posterization. Zoom in and it will disappear as the true (16bit) tones are displayed.

Suffice to say I haven't seen Tony's video, but I think we agree that compression is good, and bad, depending on what you want.

Well, unless you buy a Fuji GFX 100.