mark_herring

Answer to the question in the title: Stitching. Look at "Kremlin Sunrise" in my gallery. 8 shots from a 2Mp Canon A40, stitched in PTAssembler/Panotools, and then cleaned up in Photoshop.

 

More general answer: As with all SW, the real secret is to not be afraid to try things. It always amazes me to see people who a actually AFRAID to randomly pull down an menu and try something. With proper computer hygiene, you cant break anything.

 

My latest trick---surely not original--is to do selective burning by putting a partially opaque layer over evertyhing, and then using the eraser as a burning tool.

You can replace the waste ink pad and reset the counter. The procedure has been published in various places, Including--I think--this forum. Search on "waste ink counter".

If you enjoy keeping track of Epson annoyances, I understand they will not provide this procedure to customers.

For Mike Morgan;

 

Are you suggesting that the OP here is really my old friend George Preddy? God help us if so.

I started with a 2Mp Canon A40--hardly a bank breaker. Excellent pictures with no processing, but BETTER pictures with a few simply tweaks---the kind of thing you would do in the darkroom without even thinking about it---plus some things you could not easily do in the DR.

Hassle? NO. for snapshots, I have a routine wherein I can almost keep up with my printer (running 4" roll paper). What I do to every picture:

crop (1-3 minutes): It is the very rare picture that does not benefit by a bit of trimming. On maybe 25% of the pix, rotate while cropping so some logical feature is either horiz or vertical.

levels (2-4 minutes): Ensure that there is full black and full white, and tweak the midpoint a bit if necessary.

Unsharp mask (3-5 minutes)

Save and print

 

You can now get into digital cheaply, with excellent results, and no hassle. Buy a 3Mp P&S, preview on your TV, and get prints made at the local camera store. You'll get good results, save money, and have no hassle. You'll also miss 95% of the rewards of digital.

 

It is NO different from the film world. You can get started for $200, $2,000, or $20,000.

I do mostly cut and try.

 

One interesting reality is that a sharp picture wants different settings than a fuzzy one.

 

Starting point: amount 70, radius 1.5, threshold 1 or 2

 

First, change the amount---looking for the best increase in sharpness but without getting the halo effect. Look in an important area (eyes, hair, tree branches, etc.) Then play with the radius. First go to a lower setting and see if you can still get good results. Some pictures work well at 80/2, while others want 150/1.5.

 

I almost never change the threshold--I normally leave it at 1.

Wow....

Go to Google Groups and search on "Foveon" or "Sigma". You will find the most long-winded discussion in the history of Usenet. The principal over-zealous missionary goes under the Pseudonym of George Preddy, and is now in any number of kill-files. He is reported to have similarly terrorized an Amiga group for **six years**.

 

Most folks in rec.photo.digital have not been favorably impressed by the Foveon.

 

As for theory:

To the extent that the megapixel rating is useful, it applies to spatial resolution--ie the ability to resolve detail. (To be more precise, the number of pixels defines the **limiting** resolution.) According to what I have called the "de-facto standard, a pixel must be defined as a spatial sample that contributes to the spatial resolution. By this test, the 3 stacked sensing sites in a Foveon only count as one pixel. Thus, the Foveon is a 3.4Mp device (again--in accord with the de-facto standard).

While some reviewers have reported performance similar to 5Mp "conventional" sensors, many others argue that this is an artifact of the aliasing behavior of the device.

 

With respect to color spatial resolution, it is certainly true that the Bayer has lower resolution in color than it does for luminance. What saves the Bayer, however, is that--above a certain pixel count---We don't care! This is of course that same principle that enabled compatible color TV circa 1950.

 

Another issue is color fidelity. Many critics report inaccurate colors. I have argued that the nature of the device precludes individual tailoring of the filter band-shapes. Thus, imperfect color rendition may be a non-fixable attribute of the technology.

 

Before making the decision to propogate this in this forum, I strongly recommend the Google Groups search in the archives of rec.photo.digital

Stay in touch on this!!!---I'm in the same boat. I love the A40 and the biggest motivation for an upgrade is a longer zoom.

The 5400 and G5 are both 1/1.8 sensors, and both are 4X zoom. The biggest difference I see is that the 5400 goes down to 28mm equivalent. The G5 has the faster lens. I'd expect the image qualtiy to be similar

I'm considering the Nikon 5700: larger sensor and wider zoom. I wont miss the wide angle end, since I have started using stitching for all the wide angle stuff (where the subject most often holds still!) With the rebate, the 5700 is now under $600.

The manual should tell you, but here's a guess: If there is not a specific command to shut off the pop-up, then it probably is automatically disabled when you connect to the hot shoe. That's how I would design it, but what do I know?

Just returned from a trip to Mexico and had TWO CF cards glitched

somehow. They went thru the carry-on luggage x-ray, but otherwise

did not see any unusual environments.

 

Fortunately, recovery SW saved the pictures. I used "ImageRecall"

(http://www.flashfixers.com/) which worked flawlessly. After

recovering the pictures, I used their utility to check all my CFs.

Everything checked good, and there is no sign of any malfunction in

the camera that could have caused this.

 

Any similar experiences out there?

Whatever the algorithm, it is some sort of interpolation in the camera's processor---not "pixel-skipping". All of the original pixel values are used---just as if you down-sampled in photoshop.

In principle, the 3Mp downsampled to 2 should be be better. The reason is the Bayer filter architecture. With the most common (3-color) design, the color spatial resolution is roughly 1/3 of the luminance resolution. You can see the effect of this in color fringing at high-contrast edges.

For 2Mp and above, the effect is minor and does not really bother most people, but it is there.

Yes....

My Canon A40 (2Mp) has (mostly) replaced my two Oly stylus snapshot cameras (Not that I could ever part with the latter).

Why:

1. 2Mp is totally adequate for snaps

2. convenience

3. cost-effective: Now only about 5-10% gets printed, and that happens at home. With a roll feeder on my Epson's, I can crank out 4x5.4s at the same cost as a lab. For the custom stuff, there is no way to compare cost---what I do would be prohibitive if I paid someone.

 

And, its not just about snapshots. Add PTAssembler (www.tawbaware.com) for $50, and you have horizontal AND vertical stitching to create, for example:

http://www.photo.net/photo/2067254

 

In addition to the Oly P&Ss, I'm going to keep ONE "normal" 35mm---probably my trusty M3

<<Inspired by the thread: "The Eight Megapixel Invasion", posted on

Rec.photo.digital>>

 

Thinking about why 8Mp might be better than 5Mp, even if the noise is

higher:

 

1. When you blow up a digital image, you typically see

pixellation before you see random noise--at least in the higher

values. This means that the camera still has useful resolution** at

the sampling limit. Thus--increasing the sampling limit will---in

principle---yield higher performance

 

2. Everything else being equal, the aliasing will be reduced

with the higher sampling. This works with #1 to give better

performance at the highest spatial frequencies.

 

3. With respect to noise, two points:

 

A. For the same OVERALL sensor size and f/no, the

performance FOR A GIVEN OBJECT should be similar. See resolution

comments below.

B. Noise in digicams is most obvious at the low end, and

thus the effect on overall picture quality is not linear.

 

4. With the Bayer filter pattern, higher color fidelity will be

obtained when down-sampling an 8 Mp image to 5Mp---as opposed to

native 5Mp

 

**Resolution:

I am using a rigorous definition, wherein the ability to resolve a

detail is determined by the ratio of the imaged contrast to the noise.

 

BOTTOM LINE: What would I expect to see in a comparison of an 8Mp

digicam with a 5? Both using the same total sensor area and the same

lens.

 

--higher noise in the shadows, but the same if downsampled to the

same file size

 

--higher limiting resolution (at least for high-contrast objects)

 

--less obvious pixellation

 

--fewer color-undersampling artifacts

 

--When down-sampled to the same file size, overall image quality

slightly better, but perhaps not enough to warrant the money. For

someone like me, who is about to upgrade from 2-3 Mpixel, the price

of 8 vs 5 will very likely prove to be a bad trade.

 

When I can do this:

http://www.photo.net/photo/2067254

with 2Mp, then I am going to be in "pixel heaven" at 5

 

I hope someone will post some 8Mp images that we can use to compare

to native 5Mp

 

The REAL benefit may be the price drop for 5Mp---witness the Nikon

$150 rebate on the 5700!!!

 

-Mark

 

On 14 Feb 2004 03:00:09 EST, Jef Poskanzer <jef@acme.com>

wrote:

 

>http://www.acme.com/digicams/8mp.html

>

>Comments welcome.

>---

>Jef

>

> Jef Poskanzer jef@acme.com http://www.acme.com/jef/

> Fast. Neat. Average.

No direct experience...

 

My reading tells me that--in general--pigment has lower gamut than dye. I do not, however, know why this should be a fundamental truth. There are now LOTS of 3rd party ink choices---here are few for starters: (Use Google to get the URL)

MIS (inksupply)

Mediastreet

Lyson

Colorbat

 

It also depends on the paper. For example, Epson dye ink on premium glossy or colorlife is hard to beat. I would keep the 1280 for small stuff where you dont need archival.

A snip from the Kodak site:

 

"The longest lasting inkjet photo paper in the vast majority of cases under typical home display conditions. Lasts over 100 years when using latest inks without protection behind glass."

 

Now how on earth would I know if I was using the "latest inks"?

Ways in which digital is different---and **maybe** better---than film:

Better color accuracy

More linear response

Higher contrast (MTF) at or near the limiting resolution

Noise mainly affects dark areas whereas grain affects all levels

Lower limiting resolution---even lower in color (for Bayer)

Aliasing---esp in low end cameras

Fewer optical steps to final image

Others??

Thought problem:

Suppose you get all the resolution you need by shooting at 5 Mp. Will you have higher quality with native 5Mp, or an 8Mp downsampled to 5?

 

the one thing that might give the 8Mp the edge is the color fringing associated with the Bayer filter architecture.

 

BUT--sony added a 4th color to the soup---thus increasing the color undersampling inherent to Bayer.

Check out MIS (www.inksupply.com)

also Mediastreet, Lyson, and Pantone (they are recently moving into the inkjet market---thru a company called amazon ink in CA. Last I checked, they were not up and running.

 

My research led me to MIS, but I have only limited experience with their products.

 

BTW, there is no fundamental reason that 3rd party ink cannot be as good as the OEM. Do you always put genuine Ford parts in your Ford? Inkjet printing is very rapidly becoming generic.

Wow---expect lots of opinions! I think there are now more combos of ink (chemistry) and paper than there ever were for conventional wet darkroom.

 

The pigment inks favor matte papers, but there is the "Ultrachrome" type that does work on glossy.

 

I would browse some of the major 3rd party ink houses. They tend to have some editorial comments on various papers that work well with their inks.

MIS (www.inksupply.com),

Mediastreet,

InkJetArt,

Lyson,

 

Paper:

Red River,

Hawk Mountain,

Many others---search Google on "inkjet supplies", etc.

 

I'm using Epson Enhanced Matte---very good so far

I recommended doing at least SOME printing at home. This gives you an understanding of the whole process. For inkjet, the quality you can get at home is probably not hugely different than what you can buy.

I have searched here and on Google groups.

 

I am looking for an inexpensive icc profile editor. The only live

link that I found was to the Colorvision product (doc something) for

$99.

 

I am still struggling with choice between buying custom profiles, and

making them with a scanner-based profile package. Research so far

says that I might need to edit them regardless. Main reason now to

get the editor is just to understand better what is going on.

Be aware of a cute little "feature" of digital: Grain Aliasing.

 

Depending on the film and the scan resolution the grain can be undersampled, resulting in aliasing of the grain signature to a lower spatial frequency. Thus--depending on the parameters-- you can wind up with more--and coarser--grain. The one sure cure is to scan at the highest possible sampling rate.

There are two basic kinds of electronic / CCD noise: "read noise" and "photon" or "shot" noise.

Read noise is the noise associated with the 1st preamp that takes the signal from the detector. This is not dependent on exposure / signal level. This is what you see in dark areas of a picture.

 

Shot noise has to do with the statistical arrival of photons, and is typically expressed as a sigma equal to the square root of the number of electrons. Thus a sensor which saturates at--say--1 million electrons, will have a shot noise of 1000 electrons. You typically do not see this noise because--at the hypothetical full well, you have a signal to noise ratio of 1000.

 

Assuming that read noise dominates, then we see that digital cameras are quite different that film wrt noise. In film, the phenomenology is quite different--it is the variation in size and shpe of grains that creates the noise. It is the same at all brightness levels--maybe even HIGHER at high light level.

My 1280 is now a pigment printer---running MIS original archival ink 9about to try their "perpetual" archival ink.

 

Excellent results so far

Good point from John Houghton in re clogging and CIS. The CIS keeps ink in the head at all times with no air bubbles---this has got to be the best solution. (I have had issues installing freshly refilled carts---acted like clogs, but were probably trapped air)