LF jpgs vs LSF jpgs

LF



LSF



this is one quick proof that LSF is better than LF.

the first pic is LF, the second LSF.

both are 100% crops.

the first number of the bar code, in reallity, is 8.
it is also "8" in LSF, but in LF it is "0".

also, allmost all numbers, are much better in LSF.

ok? do you agree?

nick_webster said:

I thought the big attraction of jpgs was to remove the whole PP step, which is why I said what I did in the previous post.

Click to expand...

It's not this straightforward, and largely dependent on intent and OOC JPEG result. There are other cases. Let me try and break it down this way (and illustrate with images):

• for casual shooting and when OOC is good enough, yes, shooting JPEG results in practically no PP. I am amazed how often it happens with E-5, by the way: OOC, shot with no second thought just walking by, no PP whatsoever - and suddenly even sold a few copies.

• for shooting with PP in mind but the PP is of a nature I described in previous post. No advantage to using Raw.

Original:



PPed:



While it may be possible to come to the final image in Raw, I would be hard pressed to do it, and certainly it's much easier to do in JPEG. I can think of quite a few examples simply impossible to do in Raw.

nick_webster said:

Wasn't trying to say what people should or shouldn't do, I hope it didn't come across that way

Click to expand...

Well, a lot of comments in this kind of dialogs are being made in just that manner, as if there were just one way to do things, or objectively best way to do them... I prefer exchange in mutually-enriching manner rather than in combative, so glad to hear you concur.

nick_webster said:

I realise there is a theoretical benefit to using LSF but I'm curious to see if there are any actual examples - or is it a case of better safe than sorry ? Have you ever tested LSF against LF to see if you can see any differences ?

Click to expand...

I haven't - it simply never crossed my mind to check if there is an advantage to less-compressed file vs. more compressed. It's not like each photographer must personally check every concept ever formulated... and the fact that JPEG is a lossy format and more compression introduces more loss is well established a few eons ago... On top of that, as far as PP is concerned, such test is not simple nor quick. See my previous exchange with GB in another thread where I went into details to show why.

Yes I can see a difference in those jpgs.

Perhaps it is camera dependant as I'm just not seeing it with files from the E-3. Maybe the jpg engine is using better algorithms or it's due to the weaker AA filter of the E-5 ? Or it might be the subject matter, I'll see if I can duplicate something similar with a barcode shot and the E-3.

Thanks for taking the time to shoot/post those images,

Nick

Nick, were these processed in-camera? I've never tried using in-camera RAW file processing in any of my Olympus DSLR's. The reason for asking is I know one cannot get the equivalent of a Large Superfine JPEG using Viewer II. The best quality JPEG setting in Viewer only yields something equivalent to the in-camera Fine setting.

More alpha-numeric photos should make it more convincing. People already have idealized letters and numbers in their head to allow them to see the difference. Good white/black contrast too. Flowers, landscape, pets..etc..the observer isn't keen on what differences to detect/notice.

Peace,
John

--
http://www.flickr.com/photos/johnnyrush/sets/

"There's design intent, and then there's what the d@mn thing actually does." - Me

'The reasonable man adapts himself to the world; the unreasonable one persists in trying to adapt the world to himself. Therefore all progress depends on the unreasonable man. ' George Bernard Shaw

'Believe Nothing, no matter where you read it
or who has said it, not even if I have said it,
unless it agrees with your own reason and
your own common sense.' -Buddha

'This too shall pass.... (Hebrew advice to Solomon)
until that which passes is life itself' - me

That's the reason for these comparisons - is it worth it for jpg shooters to use the LSF, LF or can they even get away with LN ?

Looking at the results from my efforts it doesn't look as if it makes any difference for the E-3, though costas has just posted a couple of crops from the E-5 http://forums.dpreview.com/forums/read.asp?forum=1022&message=38308601 that looks as though there might be a benefit with that camera. I'm just trying to see if I can duplicate that sort of shot with my E-3 to see if that is the sort of detail that benefits from the LSF setting or it is indeed something specific to the E-5.

Nick

another example, from the same "theme":

LF



LSF



you can see clearly, that the up-side edge of the dots that make the numbers are almost all flat in LF, you can not tell the borders of each dot, while in LSF you can see the dots. In numbers "8,9,0,2", specially in "0" and "2" the up dots of the numbers are almost one body, one straight line, while in LSF you can see the dots. The only exception is number 3, which is better in LF.

This means that the smaller the size, the most jpg artefacts, which artefacts hide details.

of course, even in LSF there are artefacts, but less.

i took all pics with E-5 and 14-35, in a tripod, using the same focus (manual focus). first i shot in LSF, then in LF, and i cropped the images saving them as tiff (if i save them as jpg's, i include even more artefacts). Uploaded as tiffs, but imageshack changed them to png's.

ok? is it enough?

could you shoot just one RAW image and then convert to LSF ans LF in camera? Leave the text very small in the viewfinder. The letters/numbers should be barely discernable.

Peace,
John

--
http://www.flickr.com/photos/johnnyrush/sets/

"There's design intent, and then there's what the d@mn thing actually does." - Me

'The reasonable man adapts himself to the world; the unreasonable one persists in trying to adapt the world to himself. Therefore all progress depends on the unreasonable man. ' George Bernard Shaw

'Believe Nothing, no matter where you read it
or who has said it, not even if I have said it,
unless it agrees with your own reason and
your own common sense.' -Buddha

'This too shall pass.... (Hebrew advice to Solomon)
until that which passes is life itself' - me

pris said:

nick_webster wrote:

• for shooting with PP in mind but the PP is of a nature I described in previous post. No advantage to using Raw.

While it may be possible to come to the final image in Raw, I would be hard pressed to do it, and certainly it's much easier to do in JPEG. I can think of quite a few examples simply impossible to do in Raw.

Click to expand...

well, this is quite true.

in photoshop you can choose to open a jpg file as raw file, by using the "open as" option. With this "trick" you can make almost anything with a jpg file, but you can not save much of burned whites, which is almost true for orf files also, but you have a little more to save in the orf's.

The only real reason to shoot in raw is to squize all the possible detail you can get from your image, without having to deal with jpg artefacts. This is really helpfull in generous crops, but using a top-pro lens is a "must". If you are going to use only 4 or 5 Mp images for prints, by resizing the initial 12 Mp image (and not cropping it) then there is no need for anything like raw's and super sharpening tools for all the extra bits of small details, but also there is no need for the super light AA-filter of the E-5, nor the true-pic 5 engine. Any cheap KIT lens with a small-cheap body can do the job just fine.

My findings are.

If you shoot raw and then convert there is very little difference in LF and LSF.

But with jpg PP. Especially sharpening, artefacts start appearing earlier in LF than they do in LSF.

--
Collin

http://www.pbase.com/collinbaxter
http://collinbaxter.zenfolio.com/

Life is not measured by the number of breaths we take, but by the moments that take our breath away. (George Carlin)

Your first example was clearer, but you aren't quite following the proper test procedure. That is you should shoot a RAW and then convert that in camera altering the compression ratio each time. What you have done there is take 2 separate shots and compared them which is not quite the same. Any differences between your 2 shots may be due to factors other than just the effect of the jpg compression. I admit that there may be no difference due to taking 2 shots instead of one, but to eliminate the possibilty that the tripod moved, the light levels changed etc. you need to work from just the one RAW file and use the in camera editing facility,

Thanks for taking the time again, I do appreciate it,

Nick

Though so far I can't find any difference even at 200% on an un-PP'd ( is that a word ? ) image

NIck

You're not going all French on me ?

( subsitute generic xenophobic nationality of your choice if required - Italians are always good for a surrender - http://www.bbc.co.uk/news/world-europe-13141495 )

Nick

johnnyrush said:

could you shoot just one RAW image and then convert to LSF ans LF in camera? Leave the text very small in the viewfinder. The letters/numbers should be barely discernable.

Click to expand...

i did that, and the resoults are (more or less) the same. Well, mayby the differences are a little less obvious, but the idea is the same, you just have to see harder the details. Anyway, it is just more convinient to me to shoot 2 times, than to raw edit the file with changing of the desired jpg compression for each conversion.

the differece is small, but it is there. it is what i showed you. I also am quite sure that with the KIT lenses you can see no differece at all. That's why i posted the MTF graphs.

if you want, try something else:

shoot one raw image and make 20 LF files, and 20 LSF files, from the same raw image.

then view the LF files, in 100% crop, like a movie, with ie 3 frames per second.

you will notice many changes from frame to frame. these are the different places of the jpg artefacts, which is quite random.

Do the same, with the LSF frames. The changes from frame to frame are much much less. This is becouse there are much less jpg artefacts, due to less compression.

If you want no artefacts, try orf's... lol

You can shoot jpg and most of the time do all the processing you want to and not see any appreciable detriment compared to shooting raw. Does that mean there is no reason to shoot raw? Of course not, but maybe it isn't 'worth it' to lots of people. Even knowing that some images will show differences rather quickly.

You can open and resave jpgs over and over and over - do you see the damage being caused each time? I doubt it very much. Does that mean tiffs are useless marketing fluff?

Remember when Olympus used to offer us option of shooting directly to tiff? I think lots of Olympus unique options have gone away because they aren't offered by Canon or Nikon - ipso facto, they are stupid options. Anyone who complains about the dumbed down menu or lack of options on the gold award XZ-1 should thank dpreview for their 'help'.

I can see the differences in your last set of examples (flowers) - look at the backgrounds for noise and artifacts - increases as the compression increases. I couldn't see differences in the first examples, but that doesn't 'prove' anything except as usual - there are always compromises and no single right answer for all photographers, all images or all uses.

--
Roberto M.

nick_webster said:

Your first example was clearer, but you aren't quite following the proper test procedure. That is you should shoot a RAW and then convert that in camera altering the compression ratio each time. What you have done there is take 2 separate shots and compared them which is not quite the same. Any differences between your 2 shots may be due to factors other than just the effect of the jpg compression. I admit that there may be no difference due to taking 2 shots instead of one, but to eliminate the possibilty that the tripod moved, the light levels changed etc. you need to work from just the one RAW file and use the in camera editing facility,

Thanks for taking the time again, I do appreciate it,

Nick

Click to expand...

Nick, i am really sure that nothing changed between the 2 shots.
the tripod is a brand new manfrotto, with a big ball head, capable of 8 kilos.
i shoot with 2 seconds delay to avoid any movement of the body.
the light is artificial, no change here, inside the house.
focus is manual, no change in focus.
all other things, are completely the same. f/2.8, etc etc etc.

And, in the other hand, if there is a change between the 2 shots, it should not be in favor of the LSF file always! lol

And i get always the same resoult: the image haw better detail overall in the LSF file.

Well, this is the case.

In the other hand, when i convert the raw file to LSF and LF, for some reason i cannot be sure, the resoults are somewhat closer, but the LSF always is a step above.

The real need to know, is if you do not shoot raw, which format you will choose. I am sure that LSF is the best choise, even for postprossesing a bit. But on the other hand, if you shoot raw, there is no meaning to convert the image to jpg inside the body. You can get much much more detail by using the raw file in adobe raw, an many awasome sharpening plugins, like those from topaz labs and from photokit sharpener.

personally, i own many top-pro zuikos and i only shoot orfs. and i keep all the orfs, after all hard disks are really cheap, compaired to the top-pro zuiko collection i have! lol

happy raw shooting man!

My point was that I can't see any differences between the 3 levels of compression I used if I view them normally or print them, though if I look at > 300% I can see that each is slightly different

I started this as a way of asking if other people could see some differences that were escaping me, not to try and say one was better or worse for anyone. I didn't expect to see much, if any, difference between LSF and LF, but I've been really surprised by how very close LN files are to the 3x larger LSF files.

Hopefully other people may have learned something today, even if it was only to confirm what they already knew ...

Nick

Nick, I have something curious to show but before I do, let's see if we agree on methodology. Here is a list of steps:

1. Overlay LF and LSF files, change blending mode to Difference, add Levels adjustment layer, move the sliders almost all the way to the left to make the difference visible. Save file as diffbeforepp.jpg

2. Open both files and perform the same postprocessing on both. Overlay them, repeat all the steps from 1, save as diffafterpp.jpg.

3. Compare files visually. If they are identical or close enough, applied postrpocessing hasn't made any or much difference in quality between the two. If they are drastically different, initial compression difference has been a significant factor.

Does the above sound right to you?

nick_webster said:

You're not going all French on me ?

( subsitute generic xenophobic nationality of your choice if required - Italians are always good for a surrender - http://www.bbc.co.uk/news/world-europe-13141495 )

Nick

Click to expand...

funnily enough I'll be going to France soon with an absolute.......eccentric! So i expect his language to be quite fruitful, whenever he sees a Spitfire he shouts ' Achtung Spitfire!', and then tells everyone he hates the Germans because they bombed his chippy in the war, he's an absolute marvel for inventing aerospace composites though, ooops once again I've wandered into the wonderful garden that is OT
--
to me........to you

I'll be the first to admit my lack of PP skills, but what you suggest seems reasonable. I'd guess that as you do more PP you'll see a bigger difference between the two files. I've not got Photoshop anymore otherwise I'd try the same thing - I'm trying to stick to Aperture for my own sanity ( too much software, not enough time to learn them properly )

This is getting beyond my OP which was really looking at what Andy Westlake was asserting about LSF and LF files, but feel free to show your results. If there's enough interest it might be worthwhile if you started a new thread displaying your findings,

Nick