RAW -vs- s/mRAW -vs- CS4

[battletone]

I am heavily leaning towards a body upgrade, and right now the 7D and the D300s are the two I am looking at. But I do not feel any need to increase my megapixels...currently at 12. 18MP, and even larger RAW files, is not something I am looking forward to. Space is cheap, but it still isn't free.

I have never felt Photoshop lacked in resampling images to a smaller, or much smaller size. But what about these cameras that offer sRAW or mRAW...is there any legit reason (not including the permanent downsize) to resize in PS instead? I am talking, will the finished product be cleaner in PS? Too close to call?

 

[Flash Harry]

I think your getting confused, more MP will enable greater enlargement, and downsizing a photo can be done in any image editing program, its upsizing where the problems arise, also raw converters give options for small to large files, I suppose the sraw & mraw will be useful to togs for speed loading on to computer or if they know beforehand only smaller images will be required. H

 

[battletone]

I think your getting confused, more MP will enable greater enlargement, and downsizing a photo can be done in any image editing program, its upsizing where the problems arise, also raw converters give options for small to large files, I suppose the sraw & mraw will be useful to togs for speed loading on to computer or if they know beforehand only smaller images will be required. H

?

I realize that almost any imaging program can downsize an image. I am asking if PS produces a cleaner image when it is used to reduce a RAW image to the equivalent size of the camera producing a sRAW or mRAW.

I am asking about the cameras quality of RAW files that are smaller than full size. Both have to eliminate information....does one do it better than the other? I have no intention of shooting at 18mp from the get go. I know Photoshop produces an acceptable result in my eyes, but I would rather start with 12MP+/- from the camera if possible.

 

[Big Mike]

This is a good question.
I have no idea what the process is for an 18MP camera to give you a 12MP RAW file.

Have you Googled this?

 

[Derrel]

The desire to upgrade is understood well by me. WHat you want the camera to excel at would determine what camera would be an upgrade. I've been looking into the EOS 7D (named after a Minolta!) quite a bit lately, since its price and high MP count make it seem attractive. I've seen some nice work done by that camera under bright, OCtober texas levels---such as bird photos from Fred Miranda's web site, many shot at ISO 400 at f/7.1 at 1/2000 to 1/4000...Gulf Coast, bright days, marine environment with water and light-colored reeds reflecting like all over the place. Big deal--light levels that high do not exist where I live except for about 60 days out of the year, and only in certain locations. High "cropability", but the low-light and high ISO stuff not shot in the Gulf Coast at even 1600 looks sub-par.

I just looked a the Robgalbraith.com sneak peak of the new 12.3 MP Nikon D3s and the new 70-200 VR-II lens at the Big Apple circus dress rehearsal. The images looked excellent, shot at crappy light levels like ISO 3200 at f/2.8 at 1/400s second. Obviously, the current world leader in a High-ISO body, but fewer MP.

My point? No, I disagree with the poster above that higher MP equates to larger enlargements; what equates to higher enlargement size is higher Image Quality, meaning that better per-pixel level detail and lower per-pixel noise is what counts now, in 2009. A few years ago, we needed higher and higher MP counts, but the 12MP cameras have shown us that the full-frame 12.3 to 12.8 (Nikon, Canon original 5D) sensors can produce exceptional per-pixel level quality with a wide variety of lenses of modest specification. Large pixels (8 microns or larger) in the 5D, D3, and D700 and D3s give low noise and have ample resolving power to produce large, up-sampled images if super-large prints are needed. Even in crummy light and at elevated ISO settings. Small pixels (4.9 micron size like the 7D has) packed exceptionally densely on a smaller 1.6x FOV sensor are showing that with professional-grade lenses it is possible to make captures with high resolving abilities, but noise quickly becomes a problem as ISO levels go above medium and into the High-ISO zones.

So, the question on upgrading really is--what kind of work do you want to shoot? What light levels and what types of stuff do you want your camera to be optimized for? And, of course, Full Frame or crop-sensor? A few years ago, an "upgrade" was automatic and simple--just go with the newer body with the higher MP count. But today, Nikon and Canon have very carefully created two entirely different systems that do not compete head-to-head. The 5D and 5D Mark II do not compete with the D700 for action shooters.
The 7D is Nikon's answer to the D300s, yet the D300s is an interim place holder due for replacement in 18-24 months.

 

[battletone]

This is a good question.
I have no idea what the process is for an 18MP camera to give you a 12MP RAW file.

Have you Googled this?

Yes, and the answers are all over the board. If I had a camera that could shoot medium or small Raw files, I would just test it out myself. Some say you get less noise than full size, others say you get less IQ. None point to pixel peeping demonstration.

The sRAW on the 7D appears to be 1/2 of the pixels in each direction. Some say it "merges" the adjacent sensor pixels, but none point to anything from established "in the know" websites. I would think it would be easier to just not record the data. But maybe the merge "averages out" and gives less noise? Who knows. I don't even know how PS does it. I just want to see what it looks like.

So, the question on upgrading really is--what kind of work do you want to shoot?

Everything from family P&S to baseball games to landscapes and wildlife. Too much for one camera huh? Well my XSi is nice. I like it. For me I find it to lack very few things that I would like added.

...that said. My three areas I am tyring to improve on...

• I want a higher usable ISO. I don't need 6 digits, but the image quality at 3200 from the 7D would be nice.
• I don't need it per say, but now that I have seen how much faster my autofocus is than my previous camera, I would like it even faster.
• Weather sealing. Be it dust in AZ or rain in MN and WI, I want a decent level of sealing. A metal body doesn't hurt either.

 

[Overread17 5]

Out of interest what lenses do you have currently? Your profile lists only a 450D and a kit lens for that camera - is that all you have or have you added more?

 

[battletone]

Out of interest what lenses do you have currently? Your profile lists only a 450D and a kit lens for that camera - is that all you have or have you added more?

No I haven't added more and will not add more until I decide on the body, because if I end up with a Nikon, the Canon lens investment is a waste of money. This was my first DSLR, and was the first SLR I really got to get a feel for the functions of. Now I am getting a better idea of what I want/need and where I want to go.

Please redirect this back onto the issue of RAW and sRAW etc.

 

[musicaleCA]

AFAIK mRAW and sRAW just average more photosites to create a pixel. Ish sorta. Instead of using four photosites (two green, one red, one blue), say, in mRAW the camera instead uses a 4x4 (which would make, lesse, 8 green, 4 red, and 4 blue; this is only an example, and actual math of the number of photosites on the darn sensor will likely prove me wrong, so if you're a math geek, go suck an egg, I'm not trying to be accurate ) square of photosites with equivalent overlap. I think that's how it's handled in-camera; dont' quote me on that. (Actually, it probably drops the overlapping pixels first...hrm...)

As for image quality, dur, it's going to smush your fine details. That's a given. It doesn't reduce noise at all. What happens is the noise that was there is averaged. This can actually make it harder to remove in post, in my experience.

Where it's useful is when you're trying to save space and you know a particular shot doesn't require lots of detail. Wedding photographers, I hear, will sometimes switch to mRAW to get a shot of the cake or something else relatively trivial that won't be printed at a large size.

 

[Garbz]

Erm. Noise is random. If you average a random function centred around zero it becomes closer to zero, noise IS reduced.

However what is really the issue here? The size? battleone space is not free but it is cheap, LUDICROUSLY cheap. 1TB can be had for $100 last I remember. On a 1TB hard disk you can fit 62500 RAW files from a D200 (16MB ea). Which means if you buy 2x 1TB harddisks your camera's shutter will completely wear out before you fill the two with RAWs files, assuming you don't ever delete an image?


Here's another thought. Do you re-edit your old files? I don't, and thus I don't store RAWs since there's no point. The RAW only stores data that I can't see. My final edited and finished images are all high quality JPEG. On a 1TB harddisk for $100 you can store about 200000 images assuming 5MB.

In the last year I have taken about 10000 photos that I wanted to keep. That takes up 40GB of harddisk space. If I had saved the RAWs instead of the JPEGs it would be 160GB, which represents about $16 worth of storage on a 1TB harddisk. (or $32 in my case since I would want a redundant backup). This redundant solution for 1 years worth of shooting is now the same as 2 movie tickets, or buying 1 Bluray.


Space is essentially free! Compare the cost of the harddisk to the cost of the camera, and don't sacrifice your quality images for something so mundane.

All prices are in AUD

 

[musicaleCA]

Erm. Noise is random. If you average a random function centred around zero it becomes closer to zero, noise IS reduced.

Then why do people keep telling me it isn't! Crazies! CRAZIES ALL OF J00!!! AHHHH!!!

Either way, in my experience, you get a bit more out of the large file and reducing noise with something like Nik Define, over straight downsampling (well, mostly because you can remove the noise and THEN downsample).

Don't suppose you can shed any light on how it's actually downsampled in camera though, eh Garbz? I think I got it right...ish. Maybe...sorta...

 

[battletone]

Either way, in my experience, you get a bit more out of the large file and reducing noise with something like Nik Define, over straight downsampling (well, mostly because you can remove the noise and THEN downsample).

Well that is a valid point to consider. I hadn't thought about that.

Erm. Noise is random. If you average a random function centred around zero it becomes closer to zero, noise IS reduced.

However what is really the issue here? The size? battleone space is not free but it is cheap, LUDICROUSLY cheap. 1TB can be had for $100 last I remember. On a 1TB hard disk you can fit 62500 RAW files from a D200 (16MB ea). Which means if you buy 2x 1TB harddisks your camera's shutter will completely wear out before you fill the two with RAWs files, assuming you don't ever delete an image?


Here's another thought. Do you re-edit your old files? I don't, and thus I don't store RAWs since there's no point. The RAW only stores data that I can't see. My final edited and finished images are all high quality JPEG. On a 1TB harddisk for $100 you can store about 200000 images assuming 5MB.

In the last year I have taken about 10000 photos that I wanted to keep. That takes up 40GB of harddisk space. If I had saved the RAWs instead of the JPEGs it would be 160GB, which represents about $16 worth of storage on a 1TB harddisk. (or $32 in my case since I would want a redundant backup). This redundant solution for 1 years worth of shooting is now the same as 2 movie tickets, or buying 1 Bluray.


Space is essentially free! Compare the cost of the harddisk to the cost of the camera, and don't sacrifice your quality images for something so mundane.

All prices are in AUD

Well, the time it takes to back up my files is also a consideration, especially when I have two systems. That doubles everything. Time is money. I am a penny pincher...not in buying cheap to be cheap, but in buying what I feel I need after weighing all the factors. Then you consider that everything is digital, and will be for a good while, meaning that overtime, things will add up....not not just pics, but music and other media. So while 25% more data per picture might seem like it isn't an issue, its still compounding over time...and in 10 years, the transfer rates will be far beyond what they are now, I have no doubt, but what I currently have is already a burden in my eyes.
So when I factor in that I don't need anything over 12mp IMO, adding 25% extra it not something I take without consideration.

I cannot say I had considered editing RAWs and only keeping a JPG. I guess that is a viable option.

 

[Garbz]

Well the way I look at it, 4mpx is enough for me, until it isn't. Sure I mostly print 6x4s but if I only had a 4mpx camera that one 40" wide image on my wall wouldn't look as good.

Still on that save for JPEG point. If you're final goal is to always save a smaller file, then so be it, but don't cripple your camera right at the start for the sake of being able to transfer the files to the computer in 2min instead of 5min. One day you may actually think wow this is worth printing large, but you'll get worse results.

Time may be money, but computers multitask for a reason. Drink a cup of coffee, read your email, or come back and post here while you wait for the transfer to complete

Actually as an aside this is one of the reason I still use my really crappy old memory card reader. I get home, the card goes in my computer and Lightroom starts automatically importing, and while it does it's wash my face, raid the fridge, and then complain bitterly that the house is the mess. When that is done my computer is ready and waiting :lmao:

Then why do people keep telling me it isn't! Crazies! CRAZIES ALL OF J00!!! AHHHH!!!

Either way, in my experience, you get a bit more out of the large file and reducing noise with something like Nik Define, over straight downsampling (well, mostly because you can remove the noise and THEN downsample).

Don't suppose you can shed any light on how it's actually downsampled in camera though, eh Garbz? I think I got it right...ish. Maybe...sorta...

What who would say that. . You're definitely right though. Actual noise reduction is many times better and more useful than throwing away data. As for how it's actually downsampled? I don't know. If it's nearest neighbour method (I highly doubt it) then you'd get no noise reduction benefit. But anyway it's all academic since it quite simply shouldn't be done in camera. You don't know what you're getting

 

[Derrel]

I archive and store RAW files. Why? When I began digital photography in 2001, RAW development software was in its infancy with the Nikon D1,as was my skill in developing raw files. My computer systems and monitors back THEN were nowhere near as powerful,fast,and spacious as they are today. This held true for several years thereafter,and to an extent, is still true today. The more data one has to process, the more possibilities.

As to Garbz's point; regarding storage. I would ask why if as he says, "space is not free but it is cheap, LUDICROUSLY cheap" then why he does not walk the walk he talks and keep his RAW files??? If space is "LUDICROUSLY cheap," as he puts it, then why be content with keeping a mere 8-bit JPEG image when he could be archiving 12-bit or 14-bit (real data) raw files? Seems a bit of a disconnect. Having been at this myself since the early days of d-slr photograhy, I have seen a few absolutely MAJOR advancements in image processing software, such as the invention of D-Lighting AKA FIll-Light optimization in software, Highlight Recovery software, the Adobe Camera RAW software application, the Lightroom speed advantage landmark, and the quick-masking technologies developed by Nik Software and first used in Nikon Capture with the Control Point Technology, and now sold as Viveza; this is a technology the folks at Adobe have tried to imitate, but can not yet equal, especially in terms of ease and speed of use.

Again, if storage space is "LUDICROUSLY cheap," I would think a fellow who can afford $100 terabyte drives galore ought to see the value of saving 12- or 14-bit data, unless of course he has not been involved in digital imaging long enough to have seen how crude software and computers were nine or ten years ago, and who doesn't have first hand experience of how the sudden development of tremendously better software can breathe entirely new life into archives of older, 12-bit RAW images shot just seven or eight years ago. With 8-bit JPEGS, the majority of the information is thrown out. RAW files are your digital negatives! Keeping JPEGs only is tantamount to keeping only 4x6 color prints, but throwing away the negatives, due to lack of space to store said negatives. I mean, we have ludicrously cheap storage,right? Why not keep the negatives,and make new prints as needed?

But back to the original question about the 7D and medium and small raw files: opinions seem divided,and one observation I have seen is that from those who own both the 5D Mark II and the new 7D--the 5D-II does a better job on its reduced-size RAW captures than the 7D does on its reduced-size captures. On average, a rough generalization is that reducing in size by half reduces noise by a factor of about 1.4. But the problem really is that, at a per-pixel level, the 7D is significantly higher-noise than the 5D-II is. I understand that musicaleCA owns a couple of 7D cameras, but does not have experiecne with full-frame cameras,and he might be blinded by his new love, the 7D. But from what I have been able to glean from Fred Miranda's many dual-camera owners, the remarkable performance of the 5D-II's reduced size RAW captures is *not* carrying over to the 7D's reduced captures,especially those shot at higher ISO settings.

Applying a boatload of noise reduction to a 329 square millimeter 7D capture of just under 18MP is one thing: applying NO noise reduction to an 864 square millimeter, full-frame capture from a 12.8 MP EOS 5D or a 21.5 MP EOS 5D-II or a 24.5 MP Nikon D3x capture is an entirely different thing. We're talking about a very complex interaction between pixel size and well capacity, sensor dynamic range, lens performance, and native, straight out of camera per-pixel image quality. One thing I know: a powerful PC with ample RAM and modern,sophisticated software, with 12- or 14-bit RAW data can make a very fine image file; probably better than one the camera's on-board processor can make in 1/10th of a second at shooting time.

Based on the increase in computer and software power and capabilities over the last 5 years, my money goes to shooting RAW and KEEPING RAW data on-hand for later processing. And, this is the part that irks many APS-C advocates, I think the lens/pixel density equation has already been maxed out with current lenses with the 7D. 17.6 million tiny pixels, in my opinion, are no match for even 12.3 or 12.8 million more-than double-sized pixels. The per-pixel noise level is better with the full-frame camera, and the color response seems richer at higher ISO's with the full-frame large-pixel cameras, and the performance of average- to mid-level professional lenses like the 17-55 f/2.8 IS Canon zoom seems to have been maxed out with the 17.6 million pixels crammed onto the 1.6x sensor size.

So, to briefly re-state what I have been able to glean: 5D-II reduced size raw files seem to do well,even at higher ISO settings. But the 7D, which has the smallest pixels of any d-slr ever made, on a small sensor, which is extremely demanding of lens performance, is creating reduced size raw files that 5D Mark II owners are not thrilled with. For whatever reasons--those who are comparing a FF, higher-MP count camera from Canon with the new 1.6x camera are giving the edge not to the newer technology,but the old-school, big-block,large-displacement,normally-aspirated V8 "ride" over the newer "ride" and its small-displacement,higher-revving, fuel-injected 4-banger DOHC, 4-valves-per cylinder technology...

 

[battletone]

Well the way I look at it, 4mpx is enough for me, until it isn't. Sure I mostly print 6x4s but if I only had a 4mpx camera that one 40" wide image on my wall wouldn't look as good.

Still on that save for JPEG point. If you're final goal is to always save a smaller file, then so be it, but don't cripple your camera right at the start for the sake of being able to transfer the files to the computer in 2min instead of 5min. One day you may actually think wow this is worth printing large, but you'll get worse results.

Time may be money, but computers multitask for a reason. Drink a cup of coffee, read your email, or come back and post here while you wait for the transfer to complete

Actually as an aside this is one of the reason I still use my really crappy old memory card reader. I get home, the card goes in my computer and Lightroom starts automatically importing, and while it does it's wash my face, raid the fridge, and then complain bitterly that the house is the mess. When that is done my computer is ready and waiting :lmao:

I think we are on separate pages. 2 minutes instead of 5 minutes to transfer off of the memory card is not the issue....its the sum of all data being backed up. My current means is Firewire, which on my next build will probably become external SATA. Regardless, my current backup seems to around an hour per drive, and I have only just begun shooting in RAW. Transfer rates cannot increase fast enough IMO.