Perhaps the lens is sharper than we thought??

Started 11 months ago | Discussions thread
OP ANAYV Forum Pro • Posts: 19,940
Re: Perhaps the lens is sharper than we thought??

Dak on cam wrote:

ANAYV wrote:

Dak on cam wrote:

The camera was told at -2.3EV and ISO80 not to worry about lack of light and the unavoidable noise was then flattened along with detail.

Detail didn't see that compromised.

Here is a crop from the sharpest part in the plumage.

NR was controllable, since RAW.

Sure, but you cannot kill noise independently from detail. And the image has been processed in a manner where no apparent noise is visible in spite of being severely underexposed on a small-sensor cam.

For a 1/60th shutter at 1200mm , handheld.

I see enough detail in this shot (perhaps more than most other shots from FZ80 with this subject...only now with much less noise!

My FZ80 just improved, lol.

I could try and use less PRIME NR and see if any more detail can be shown.

It's always a balance between the two.

I just dislike the in camera jpeg approach, which shows LESS details and MORE noise than what's possible with DXO.

I'm still a rookie using hopefully the best is yet to come.

Do you want the RAW file?

Not really. For one thing, it is surprisingly hard to pass raw files through dpreview (basically, you have to agree on some upload service or exchange Email addresses via private messages). For another, I pretty much know that I would produce either a noisy or a smudged image from the given raw material. The funny thing is that on something like fine-grained plumage, even random luminosity noise added to a smudged base creates a sharper, more detailed impression even though it's complete fake.

True. Not for Chroma type noise ,  but yes for luminous type noise.

So whatever processing I'd add, it would not really prove a whole lot. At any rate, I think that less noise reduction would likely work better for this image, but certainly not in the same manner than more light could have worked.

No option for more light. I will try using higher ISO...but don't expect a huge difference.

But will try. Thanks for your input !

Also will try less NR.

And in this case, I'd not even blame the camera. Next time I'd try cranking the ISO up instead of lowering EV.

I will try this.

You get the same kind of light to the sensor, but the camera tries harder to help with preserving detail.

I hope so. I know with jpegs it doesn't. Too much NR.

Ok, some JPEG processors might interpret the ISO level as a blanket noise reduction permission. Then the -EV would be sort of a workaround for that. With raw, it might not be needed.

Yeah. RAW to the rescue here, me thinks.

When talking about raw, the principal advantage of -EV over raising ISO on sensors with a small dynamic range is that -EV is less prone to blown highlights. Basically you keep more of the dynamic range for the high end while telling the camera not to bother with the low end.

It smooths them out too much for my taste, and still some noise is there.

In-camera JPEG processing tends to be fine when you are not struggling to make ends meet with regard to noise/light. Then one can make do with what amounts to a fixed "sweet spot" of processing rather than finding the best one for each individual picture.

I don't mind processing individual images.

Yes, this will add noise, but less noise than pushing underexposure will, so you can go somewhat lighter on the noise reduction in post-processing.

Thanks will try.

Many sensors today are ISO invariant.

That tends to be more the case with larger sensors I think.

I think your right.

These don't show any benefit in using higher ISO, and no noise penalty by under exposing.

Underexposing always incurs a noise penalty. ISO invariant cameras create the same raw noise levels (and images) for -EV as for +ISO. However, in-camera JPEG processing (and more likely than not, the defaults from a typical raw processor as well unless you engage autoexposure mode) then keeps the image darker for -EV and scales it up again for +ISO.

Here is an image where I accidentally kept a year 2005 CMOS sensor camera (less sensitive than CCD at that time, and ISO400 tends to look rather noisy already on this cam) at ISO3200 and the flash was incapable of turning itself down that much, so it's at least +2EV (I think I used -2EV in the raw processing then).


Even pixel-peeping, the noise you'd expect just isn't there. That's the kind of reaction you expect from "ISO insensitive" (which that camera isn't actually): what the raw data before processing ends up does not depend on the ISO setting but on the light arriving on the sensor. In that case, +EV was almost as good as -ISO. But for a small sensor, leaving at least the lowest ISO ranges alone when the light is insufficient might make a difference. It is mostly the low ISO range of small sensors where "ISO variance" is a thing.

I see.

Thanks for your knowledge and input.


Post (hide subjects) Posted by
Keyboard shortcuts:
FForum PPrevious NNext WNext unread UUpvote SSubscribe RReply QQuote BBookmark MMy threads
Color scheme? Blue / Yellow