Friday, April 30, 2010

HS10 Review Part 12 – Dynamic Range Revisited

My regular readers will know by now that I like to pound certain points home. Sorry :-)

One thing that is a bit disconcerting after shooting for so long with EXR technology is the lack of forgiveness in the images from the HS10. You have to perfectly nail the exposure, and even then, you are likely to blow highlights sky high.

Some highlights cannot be saved no matter which technology you use. But when there is a chance, I like to save the highlights. It simply removes one unnecessary jarring element from the image, and why would I not want to do that?

So … a pair of images shot in the back yard, from about the same spot as the comparison shot in part 11, but facing 180 degrees from those images.

I am shooting a small branch of my Buffalo Berry, a tall bush with 1 inch thorns that could kill in my opinion. I hate this thing, but right now it has to stay because it is roped to my fence, holding the fence up …

About 6 feet behind the branch is the fence, and it is in shadow and is thrown nicely out of focus by both cameras, which are shot at the F70EXR’s maximum focal length of 270mm equivalence.





First off, why the contrast difference? Well, despite using a setting of “soft” for both color and contrast, the HS10 shoots higher contrast than the F70EXR. Its use of binning for dynamic range extensions definitely show up on images like these.

So I had to use more contrast for the HS10, and I left the F70EXR looking softer. And before you say that you prefer the higher contrast look og the HS10, I must point out that the F70EXR has plenty of room to increase contrast, while the HS10 does not have as much room to decrease it. Not a big difference, but remember that as we go on.

Details on the leaves and branch are similar. Good texture in both images.

So what is the big difference?

Look at the leaves on top of the branch where the dappled sun strikes full force. With the first image, the leaves are burnt to pure white. There is no detail and no color. A fairly jarring element in this image.

On the second image, you see that the EXR binning has saved all but the top right portion of one leaf. There is a bit of color and the tonal transitions are much less jarring. It’s subtle, but it’s the whole point of extending dynamic range. Retain color where other cameras cannot.

And remember that I shot both camera at –0.67EV, which saves more highlights than normal. Had I shot at 0EV, we would be looking at serious burnout everywhere.


drpankajshukla said...

Hi Kim !
I find that the HS10 Jpegs images have lost detail evn at R234 and R7 areas in the same image [so an image with high contrast ]An inablity to handle high DR is expected with the small sensors !But am I doing any thing wrong here ?[I mean apart from selecting a high DR scene to photograph with this instrument !]

As I turn to RAW The silky pix software help pages [in the section on preserving highlights ] mentions that one is expected to shoot by metering for the highlights ....!

I..I..I mean I thought that one is expected to meter for the highlights when shooting Jpegs and later bring up details in the shadows during pping ! while one should meter for the shadows when shooting RAW [shooting RAW to the right of the Histogram]in the high contrast scenes and later pull back the shadows !

So ..what gives ? I mean have I got my basics all wrong or did the Silkypix software manual make an error ?
Or is it just that the manual is indicating the behaviour of the cam [since it tends to have lost shadow detail as well as highlight detail pretty early in a high contrast scene ?] and advising the user to avoid shooting the RAW to the right of the histogram?

I am apologetic for the rather long query .. but I hope I am not sounding confused !
I will certainly appreciate an explanative response fron u since u have been able to gauge the instrument and its limitations fairly accurately !

Kim Letkeman said...

drpankajshukla: When you absolutely must preserve highlights, metering for them is the only way. If, for example, you spot metered them (and the spot meter was accurate) then you could set compensation at about +1.5 to ensure that they are very bright, but not too bright to be lost. That would, of course, since your shadows very, very deep. On small sensors like this, shadows might actually be unrecoverable.

As for the jpeg versus RAW difference ... metering for highlights works well in both cases. The smaller the sensor, the more important it is to be precise where you put the highlights if you want to preserve detail there. But, the more likely it is that you will block the shadows then. This is the perennial issue with smaller sensors, and is why EXR technology has been such a boon.

RAW allows much more latitude on large sensors, pulling back highlights from a long way into burnt and raising shadows that are blocked while retaining some detail. But on a 1/2.3" sensor, that latitude is pretty small. I'm betting that this is why Silkypix says to meter for highlights. Blocked shadows are often more tolerable than burnt highlights to most people.

That's the best I can do with this one.

drpankajshukla said...

Thanks Kim for your advise !I think u were very precise and clear ! Your response helped !
I have a print related query --I am placing it here for want of a better location !

While preparing a image file[10megapixel jpeg] for printing -making a 8 by 10 size print how much the file should be in MBs ? I mean I am getting pped files as small as 780kb or as large as 2MB depending how I ppthem ! So will the final print quality differ between these two options of the same file ?

Kim Letkeman said...

drpankajshukla: Printing is a bit of a black art ... but the simplest way is to create a copy of your image after processing it and at full size, then crop it to match your output medium, in this case you want 4x5 as your aspect ratio. Next, resize using either bicubic in photoshop or elements, or lanczos in something like irfanview (which is free) to get a nice sized image at exactly the output pixels per inch you need for the printer. That would be either 300ppi for Costco or Walmart Fuji Frontier printers, or for Canon or HP ink jets, or 360ppi for Epson ink jets. So final size for the former is 2400x3000 pixels and for he latter (Epson) is 2880x3600. Finally, sharpen the output so it looks a bit "crispy" at 50% or 100% on screen. Use settings for unsharp mask somewhere around amount=150, radius=1, threshold=1 to start with and tweak radius and amount until it looks ok. Save as jpeg at around 80% or better quality. That will make a fairly big file, but you don't necessarily have to keep the file if you only want one print. But keep it if there is a chance you will reprint.

drpankajshukla said...

Thanks again for the details on printing !
I will be grateful if u can addresss one more query [for now ]!
I am placing it here since I could not find your page on postprocessing !
Is there some way in post processsing that one can compensate for the image degradation that occurs when we use the digital zoom feature during image capture --HS10 offers this with --Instant zoom and zoom bracketing modes !
Thanks in advance !

Kim Letkeman said...

Digital zoom will cause the image to be softer than normal. Possibly much softer because any motion blur will be magnified, so you will be dealing with larger movement.

The only compensation you can perform is local contrast and high radius sharpening work. I am not an expert in all of the different techniques, but you will find some of them effective, especially when the image is targeted for the web. Sometimes, just downsizing is enough.

drpankajshukla said...

Thank u Kim for your responses !