BrianHalcYoN
Leading Member
You are just confirming that I am not crazy.
--
D90, D60
Nikon 18-55, 55-200, SB-600, 10.5/2.8 Fish, 105/2.8VR; Sigma 30/1.4
--
D90, D60
Nikon 18-55, 55-200, SB-600, 10.5/2.8 Fish, 105/2.8VR; Sigma 30/1.4
-
Welcome to the new forums! Please read this first. For known issues we are working to resolve, click here.
IQ: JPG vs straight RAW-JPG convert
- Thread starter eNo
- Start date
Supr X
Veteran Member
Well, comparing an out-of-camera(native) jpg with a ViewNX NEF-to-jpg converted image . . .
:|
Sorry, probably a slightly different way of saying the same thing.
--
David~
WSSA Member #90
. . . shoot like there's no film in the thing!
eNo
Veteran Member
Well, don't take this too far. If all you want to do is convert RAW-to-JPG in ViewNX, then it would appear there isn't much to be gained (still digging into that, though). However, if you need to adjust exposure or white balance, doing it on a 12-bit or 14-bit image will more often than not come out far ahead of the compressed 8-bit JPG data. In fact, except for being in a pinch, I would not recommend WB adjustments in JPG format at all.
~~~~~~~~~~~~~~
The Rule of Thirds is meant to be broken, but only 1/3
of the time.
D80/D90 gallery: http://esfotoclix.com
Photo blog: http://esfotoclix.com/blog1
Supr X
Veteran Member
PhotoLeon1935
Veteran Member
Here is the rest of the story......
I am surprised that the two look so much alike! Many on this thread do not remember Iliah Borg who was banned more than once. His rants were about using floating point math to de-mosaic the raw image. Most raw programs used integer math, which he considered inferior to real scientific math.
The Nikon cameras use a RISC type processor (Reduced Instruction Set Computer). They reduce some instructions but they have special hardware instructions that can keep track of input and output pointers and do table look-ups all in a single cycle and sometimes read out more than one row of image values simultaneously. The trick is that binary math is ideally suited for special shortcuts which allow false multiplication by using matrices of prerecorded tables to convert simply by having the value of light being read by a self indexing pointer to the tone curve and automatically write the new multiplied value for jpg faster than you could have written the raw value to the card.
Some of the techniques Borg wanted to improve on had to do with considering more that just the three color sensors used to make up a pixel. He wanted to consider other adjacent pixels to avoid moire' and improve native sharpness in spite of the Anti Alias filter. The main complaint of what he produced was that it takes a long time for a business computer to do these scientific things using the standard instruction set.
Nikon View is designed to work on a business computer so it also takes a while to process the image using real multiplication. The questions I have may only be answered by the engineers that created the latest RISC processor, but eNo may be able to give us some confidence that shortcut binary math very closely approximates the more involved real math.
It may interest some to know that the curves stored into Picture Controls only store 64 values for each color. I can tell by the output values that at least 255 values are stored in the conversion table and it may be 512 or more in order to handle round off in the conversion between 4,000 values to 255.
My advice to eNo is to question everything until we know the truth. Right on, eNo! I would like to see your results. My attention span is too short to do this myself.
Leon
--
http://www.leongoodman.com/balance
http://www.leongoodman.com/expose
http://www.leongoodman.com/d70focusnew.html
http://www.pbase.com/photoleon
http://www.leongoodman.com
eNo
Veteran Member
Leon - I'm concocting a couple of test cases that may expose the weakness in the in-camera conversion -- or not. Case 1: high-dynamic range image. Case 2: Wide range of tonality from dark to light in one or two colors. The 1st case should be easy to set up. The second, I'm not so sure about.
~~~~~~~~~~~~~~
The Rule of Thirds is meant to be broken, but only 1/3
of the time.
D80/D90 gallery: http://esfotoclix.com
Photo blog: http://esfotoclix.com/blog1
PhotoLeon1935
Veteran Member
It has always been a puzzle to me, and maybe this is the time to find the reason, but I can never imagine how a smooth blue sky similar to the one in your signature picture gets to be posterized and blotchy. It has high values and should not show noise or steps. Could it be that the fundamental curves as I have noticed before having only 64 unique values are not effectively curved out to be smooth? Could that perfect gradient in nature just be the perfect test target to show the weakness in shortcut math? Maybe new car fenders would be a good test target if you can't get back to that lake.
--
http://www.leongoodman.com/balance
http://www.leongoodman.com/expose
http://www.leongoodman.com/d70focusnew.html
http://www.pbase.com/photoleon
http://www.leongoodman.com
WineO
Veteran Member
I applaud eNo and PhotoLeon for there technical knowledge and eNo's tenacity in testing and examination of the results. However, as the maths is certainly beyond my grasp I take a somewhat different view. It may be that I am no sufficiently good a Photographer to get consistently good results from Jpeg straight out of the Camera but I prefer to think it is the variation in the type of shots I take where some are out of the cameras capability and compromise is needed. You just can't beat Raw whatever your Raw converter is in this sort of circumstance. Take an indoor pic in mixed lighting - the tweaked Raw image result will always be superior to Jpeg. Jpeg is good for where you have lots of shots in similar conditions but where you have the time and want the best out of your images Raw offers far more leeway and non destructive processing options so why go to Jpeg when it may cause you to mess-up an important shot that could have been great if you had shot Raw?
Claude
Claude
BrianHalcYoN
Leading Member
I think if I were printing on large formats, RAW would be the only format for me. Not there yet, though I know it is inevitable. The hard to believe part is that I haven't ever been that far off when shooting by shooting a lot. If I was far enough off, I have tried to figure out what I missed. I see too many comments about RAW "saving" the shot. I am not saying that you shouldn't shoot in RAW, I just try to shoot to minimize post-processing, making post-processing a RAW shot more valuable, more detail oriented.
--
D90, D60
Nikon 18-55, 55-200, SB-600, 10.5/2.8 Fish, 105/2.8VR; Sigma 30/1.4
--
D90, D60
Nikon 18-55, 55-200, SB-600, 10.5/2.8 Fish, 105/2.8VR; Sigma 30/1.4
Kiwi Rik
Veteran Member
Hi eno...
I dont shoot RAW...and never seen a RAW shot from my D90...not even loaded ViewNX.
However I would like to see a comparison of your efforts..............could you shoot RAW+jpg fine...then convert the RAW to jpg with no adjustments.....
Crop the same section at 100% from both...say at 1280 wide..and post the converted RAW file, plus the OOC jpg for comparison...
Then do the same with a converted RAW, but this time with all the adjustments required to get the best out of it...that is WB, sharpening etc....And then compare your efforts, plus the untouched OOC jpg by using your best "picture control" parameters...
Much appreciated
--
Cheers
Rik
I dont shoot RAW...and never seen a RAW shot from my D90...not even loaded ViewNX.
However I would like to see a comparison of your efforts..............could you shoot RAW+jpg fine...then convert the RAW to jpg with no adjustments.....
Crop the same section at 100% from both...say at 1280 wide..and post the converted RAW file, plus the OOC jpg for comparison...
Then do the same with a converted RAW, but this time with all the adjustments required to get the best out of it...that is WB, sharpening etc....And then compare your efforts, plus the untouched OOC jpg by using your best "picture control" parameters...
Much appreciated
--
Cheers
Rik
TheronFamily
Veteran Member
eNo wrote:
morning
The real question here is why you stated that you;
Clear?
morning
I full understand your process, and yet I query as to why you then even raised this thread at all? I said that the jpeg and jpeg from nef, IF converted with ViewNX will give the same result.
The real question here is why you stated that you;
This alone begs me to ask the question, WHO/HOW did they (side by side comparisons) get to THEIR (obviously different from your) results?
Clear?
TheronFamily
Veteran Member
BUT, the point is he is comparing HIS results to SOMEONE elses' (I could've sworn I've seen comparisons) without actually linking us to THOSE results.
In effect, his (eNos') experiment just proves that a nikon converted nef vs a nikon generated jpeg is indeed similar (I have said that before too).
This is the point of this thread as I see it, nay?
TheronFamily
Veteran Member
the truth of raw lies not so much in that it can save the shot. That's just an added bonus.
It's that your tonal range capabilities are far, far, more flexible.
if that does not matter, jpeg is fine erhm
--
Kevin - Photos 'n Prose: http://the-photo-muse.blogspot.com/
WSSA member #104
DPR on pic posting ...
http://forums.dpreview.com/forums/read.asp?forum=1034&message=30968734&q=simon&qf=m
It's that your tonal range capabilities are far, far, more flexible.
if that does not matter, jpeg is fine erhm
--
Kevin - Photos 'n Prose: http://the-photo-muse.blogspot.com/
WSSA member #104
DPR on pic posting ...
http://forums.dpreview.com/forums/read.asp?forum=1034&message=30968734&q=simon&qf=m
Supr X
Veteran Member
Wow, i got lost in the depths of this then, i guess i thought he was just comparing his own in-camera produced jpg with the same NEF converted to jpg by ViewNX.
. . . but i defer to anyone else's more thorough reading of this thread!
Happy Saturday to you, sir!
--
David~
WSSA Member #90
. . . shoot like there's no film in the thing!
WilsonEPhillips
Veteran Member
Isn't ViewNX using the same camera settings to produce the JPG? If so, then they should be identical. Am I missing something here?
Thomas Comerford
Veteran Member
Nikon's software applies your settings, just like the camera, but the actual processing isn't quite identical. The raw converter usually produces a more "crisp" image than the in-camera processing, although I think recent cameras have narrowed the gap somewhat.
--
http://www.pixelfixer.org
Brooks P
Veteran Member
Reading down through this thread I kept thinking of something that I read that stated that the in-camera JPG and the converted NEF to JPG, even when done buy NX, NX2, or ViewNX can differ due to the level of precision in the math between the computer and the camera. Leon touched upon this earlier and when he mentioned Borg I thought the article very likely was written by one or the other of the Borg’s.
The article, and no I can’t find it now and I have no idea where to even look, did state that the difference would be extremely minor, and thus not detectable most of the time. But, there remains the possibility that in some images the data would be such that the level of precision would alter the final result to the extent that a person could see a difference between the two images.
Years ago I ran into a situation where I was trying to get a new General Ledger program approved for a large hotel chain and the auditing firm kept getting different results. The difference wasn’t big, single figure dollar amounts when dealing with millions of dollars, but the auditors felt it was significant. The hotel chain’s General Ledger program was written in RPG-II and the auditing firm’s program was written in another programming language and IBM's engineers eventually confirmed that the difference in results was due entirely to the level of precision that lead to rounding differences between the two programming languages.
--
Brooks
http://bmiddleton.smugmug.com/
The article, and no I can’t find it now and I have no idea where to even look, did state that the difference would be extremely minor, and thus not detectable most of the time. But, there remains the possibility that in some images the data would be such that the level of precision would alter the final result to the extent that a person could see a difference between the two images.
Years ago I ran into a situation where I was trying to get a new General Ledger program approved for a large hotel chain and the auditing firm kept getting different results. The difference wasn’t big, single figure dollar amounts when dealing with millions of dollars, but the auditors felt it was significant. The hotel chain’s General Ledger program was written in RPG-II and the auditing firm’s program was written in another programming language and IBM's engineers eventually confirmed that the difference in results was due entirely to the level of precision that lead to rounding differences between the two programming languages.
--
Brooks
http://bmiddleton.smugmug.com/
TheronFamily
Veteran Member
lol
I have a problem there, being too thorough
Had a good saturday btw
--
Kevin - Photos 'n Prose: http://the-photo-muse.blogspot.com/
WSSA member #104
DPR on pic posting ...
http://forums.dpreview.com/forums/read.asp?forum=1034&message=30968734&q=simon&qf=m
I have a problem there, being too thorough
Had a good saturday btw
--
Kevin - Photos 'n Prose: http://the-photo-muse.blogspot.com/
WSSA member #104
DPR on pic posting ...
http://forums.dpreview.com/forums/read.asp?forum=1034&message=30968734&q=simon&qf=m
toomanycanons
Veteran Member
I second this request. I only shoot JPEGs as well. I've read enough comments from "RAW only" shooters to appreciate their belief that, with the correct PP, RAW will "always" result in a "superior" image. The last poster on this thread (except for me) is one of those RAW shooters. I'm not going to say that their beliefs are just that, beliefs (opinions for their own reasons) but you know how it is in the photographic world. Opinions for some are fact.
I'm asking about the degree of "superiorness" that a RAW shot taken side by side with a JPEG under decent bright light outdoor conditions would exhibit. I mean, no matter how nice one of my JPEGs will look is it just a RAW fact that it would look better if taken in RAW format. No questions about, always look better after the correct amount of PP?
I'm not asking about any other RAW "advantages", just whether it's not just opinion but FACT that a RAW image if PP'd correctly will always be superior? And by "superior", not just by small degrees (maybe not justifying the extra work put into the PP) but a marked very noticeable improvement?
toomanycanons
Veteran Member
My reply above should have also included: how much work is involved getting a RAW image to look at least as good as a JPEG---then how much involved in getting it to look better? Does each pic need its own unique bit of workflow or can you batch edit multiple pics with the same parameters just to get them to OOC JPEG quality. Sorry if I'm offending RAW shooters with my noob questions!
Similar threads
Keyboard shortcuts
- f
- Forum