Film Base Color Compensation

On Fri, 2 Jul 2010 19:28:17 +0400, Anton Shepelev wrote:

Hello all,

I got a question connected with the digital postpro-
cessing of color film scans.

Popular scanning software like VueScan offers what
is called "film base color correction", which, for color films, is used to compensate for the inherent
non-white color of the transparent film.

Here I’d like to ask: What exactly does one need to
do to compensate for film base color? Is it enough
to shift the white point to transparent area of the
film or not? Or, maybe, it is some non-uniform color
transformation depending on brightness?

I am asking this because simply making a transparent
(not exposed) sections of the film grey (almost
white in fact, to get the shadows on the positive
almost black) does not seem to yield the right col-
ors on the whole scan…

Thanks in advance,
Anton

It would help my answer if I knew in what way the color is not right – is it too light or dark, or is there an overall color cast?

In general, setting the white point to eliminate the orange mask is the most important step, but it’s also necessary to set the black point, and to adjust the mid-tone brightness and color. This can be done with a set of curves that you can save for use on other negs from the same roll. There are systematic ways to accomplish this.
—
Mike Russell – http://www.curvemeister.com

Mike Russel:

It would help my answer if I knew in what way the
color is not right – is it too light or dark, or
is there an overall color cast?

Thank you for the reply, Mike.

Light/dark issues can be dealt with using curves, I
know that, but colour is the problem for me.

In general, setting the white point to eliminate
the orange mask is the most important step, but
it’s also necessary to set the black point, and to
adjust the mid-tone brightness and color. This
can be done with a set of curves that you can save
for use on other negs from the same roll. There
are systematic ways to accomplish this.

I suppose this answers my question. I was setting
the colour (but not the brightness) of the white
point to the colour of the trasparent section of the
negative and hoped that all the other colours would
become correct and the only remaining thing to do
would be to adjust brightness levels using bright-
ness/contrast or curves (general, not per-channel).
But quite often I didn’t like the results and manu-
ally adjusted colours (temperature/hue) in a way
that the transparent negative no longer was grey.

So I thought that maybe some other kind of colour
substraction was needed. I don’t know the correct
term for it, but I was thinking about a non-linear
functional dependence of the white point on the
brightness, so that not only black and white points
but all the grey points were needed to define it.

This idea came to me when I found that the scans of
my very old slides exhibit a corellation between a-
channel and brightness which I don’t know how to fix
using the standard tools…

Anton

In article <i0o1j8$itl$>,
Anton Shepelev <{oogle]mail.com> wrote:

Mike Russel:

It would help my answer if I knew in what way the
color is not right – is it too light or dark, or
is there an overall color cast?

Thank you for the reply, Mike.
[…]
This idea came to me when I found that the scans of
my very old slides exhibit a corellation between a-
channel and brightness which I don’t know how to fix
using the standard tools…

Anton:

I do not know if Mike showed his website in his reply, but many answers can be found there.

See: http://www.curvemeister.com/index.html

Mike is the master of the field, and very generous with his knowledge.

On Sat, 3 Jul 2010 23:04:08 +0400, Anton Shepelev wrote:

Mike Russel:

It would help my answer if I knew in what way the
color is not right – is it too light or dark, or
is there an overall color cast?

Thank you for the reply, Mike.

Light/dark issues can be dealt with using curves, I
know that, but colour is the problem for me.

In general, setting the white point to eliminate
the orange mask is the most important step, but
it’s also necessary to set the black point, and to
adjust the mid-tone brightness and color. This
can be done with a set of curves that you can save
for use on other negs from the same roll. There
are systematic ways to accomplish this.

I suppose this answers my question. I was setting
the colour (but not the brightness) of the white
point to the colour of the trasparent section of the
negative and hoped that all the other colours would
become correct and the only remaining thing to do
would be to adjust brightness levels using bright-
ness/contrast or curves (general, not per-channel).
But quite often I didn’t like the results and manu-
ally adjusted colours (temperature/hue) in a way
that the transparent negative no longer was grey.

So I thought that maybe some other kind of colour
substraction was needed. I don’t know the correct
term for it, but I was thinking about a non-linear
functional dependence of the white point on the
brightness, so that not only black and white points
but all the grey points were needed to define it.

This idea came to me when I found that the scans of
my very old slides exhibit a corellation between a-
channel and brightness which I don’t know how to fix
using the standard tools…

Anton

Here’s a procedure that should get you pretty close:

1) scan a negative with pure black and pure white. An exposed bit of film leader usually works well.
2) open curves (or levels)
3) click the black eyedropper on the pure black part of the neg
4) click the white eyedropper on the clear orange part of the neg
5) invert the image by flipping the ends of the RGB curve, so that the curve slants the opposite of what it was before. For example, if your RGB curve originally went from the lower left corner to the upper right corner, it should now go from the upper left corner to the lower right.
6) adjust the middle of the RGB curve slightly to get the overall
brightness you want.
7) save this curve for the next image, or make it into an action and attach it to a function key.

If you use levels instead of curves, invert your image by swapping the white and black triangles. Inverting the image and using auto-levels can also get a good result, but it will not be as dependable as the above procedure.

Thanks, BTW, to John Stafford for his kind words 🙂
—
Mike Russell – http://www.curvemeister.com

Anton Shepelev <{oogle]mail.com> found these unused words:

This idea came to me when I found that the scans of
my very old slides exhibit a corellation between a-
channel and brightness which I don’t know how to fix
using the standard tools…

Anton

"very old slides" suffer colour [dye] changes. Each ‘batch’ may exhibit differing amounts and direction of drift, though usually it’s towards a reddish hue for one brand, greenish-blue for another.

Yoou’ll have to develop and save a set of corrections [curves] that you can test [load] but probably each slide will take some slight tweaking.

Mike Russel:

Here’s a procedure that should get you pretty
close: [snip]

Thank you, Mike.

I was using Lightroom which is much more limited and
with more complicated tools (Exposure and Blacks)
instead of Levels, and it’s Auto-Tone function,
although reported to do the same as Auto-Levels in
Photoshow, in fact often yields a lot of clipping.

Now I will try Photoshop or another program with
decent Curves.

PS: Also skip below to my reply to sir F. A. Rien
for some examples that I couldn’t fix in Light-
room.

John Stafford:

I do not know if Mike showed his website in his
reply, but many answers can be found there.

Thanks for pointing.

Sir F. A. Rien:

"very old slides" suffer colour [dye] changes. Each ‘batch’ may exhibit differing amounts and
direction of drift, though usually it’s towards a
reddish hue for one brand, greenish-blue for
another.

You mean manually finding a curve for each color
channel? I shall try it. BTW, here are some exam-
ples:

1. http://img15.imageshack.us/img15/4029/93569140.jpg

2. http://img695.imageshack.us/img695/7707/64327661.jpg

3. http://img16.imageshack.us/img16/977/48563022.jpg

Anton

On Sun, 4 Jul 2010 22:43:55 +0400, Anton Shepelev wrote:

Mike Russel:

Here’s a procedure that should get you pretty
close: [snip]

Thank you, Mike.

I was using Lightroom which is much more limited and
with more complicated tools (Exposure and Blacks)
instead of Levels, and it’s Auto-Tone function,
although reported to do the same as Auto-Levels in
Photoshow, in fact often yields a lot of clipping.

Now I will try Photoshop or another program with
decent Curves.

PS: Also skip below to my reply to sir F. A. Rien
for some examples that I couldn’t fix in Light-
room.

John Stafford:

I do not know if Mike showed his website in his
reply, but many answers can be found there.

Thanks for pointing.

Sir F. A. Rien:

"very old slides" suffer colour [dye] changes. Each ‘batch’ may exhibit differing amounts and
direction of drift, though usually it’s towards a
reddish hue for one brand, greenish-blue for
another.

You mean manually finding a curve for each color
channel? I shall try it. BTW, here are some exam-
ples:

1. http://img15.imageshack.us/img15/4029/93569140.jpg

2. http://img695.imageshack.us/img695/7707/64327661.jpg
3. http://img16.imageshack.us/img16/977/48563022.jpg

Anton

In Photoshop, try using Lab mode, paying particular attention to the endpoints of the a and b curves, to fix faded slides.
—
Mike Russell – http://www.curvemeister.com

Anton Shepelev <{oogle]mail.com> found these unused words:

Mike Russel:

Here’s a procedure that should get you pretty
close: [snip]

Thank you, Mike.

I was using Lightroom which is much more limited and
with more complicated tools (Exposure and Blacks)
instead of Levels, and it’s Auto-Tone function,
although reported to do the same as Auto-Levels in
Photoshow, in fact often yields a lot of clipping.

Quite, the whites are clipped and the blacks compressed. in the horse image

Now I will try Photoshop or another program with
decent Curves.

PS: Also skip below to my reply to sir F. A. Rien
for some examples that I couldn’t fix in Light-
room.

John Stafford:

I do not know if Mike showed his website in his
reply, but many answers can be found there.

Thanks for pointing.

Sir F. A. Rien:

"very old slides" suffer colour [dye] changes. Each ‘batch’ may exhibit differing amounts and
direction of drift, though usually it’s towards a
reddish hue for one brand, greenish-blue for
another.

You mean manually finding a curve for each color
channel? I shall try it. BTW, here are some exam-
ples:

The first Image is cropped so that you have no film information to work with. You can crop ‘final’ later.

I scan to include 100% of the film chip. whatever the ‘auto’ sets for the white and black, I expand to include 10% more range, if possible. If the result still shows compressed blacks or blocked whites, re-do.

In some scanners if the blacks compress, you need to -=reduce=- the exposure. It’s weird, but the exposue setting is on the negative and then it’s ‘flipped’ to positive.

On the horse, red channel blacks moved to nearly 50%, blue to about 25% There’s not point of grey that I can determine so I left that. Lighting also had a lot to do with it, being a showcase and some unknown type of artificial lighting.

On the coach, I set white to the perforation ‘hole’ – that would be the pure light passing through in the scanner. I set black to the lowest luminance reading I could find in the dark areas of the exhibit to the left. thinking that the wheel treads were as close to grey, I used that for the starting mid balance.

The ceiling appeared too blue, so I reset just the blue channel to midtone using the darkest part of the shadow in the arch.

Again, it’s not 100% film stock, but lighting also.

From here, personal knowledge and slight tweaks might make you happier.

I generally try the color balance selections on the RGB, but as with the last, can ‘trim’ with an individual channel.

1. http://img15.imageshack.us/img15/4029/93569140.jpg

ttrr.org/scans/93569140_clrbal.jpg

2. http://img695.imageshack.us/img695/7707/64327661.jpg

ttrr.org/scans/64327661_clrbal.jpg

3. http://img16.imageshack.us/img16/977/48563022.jpg

Sorry for the smaller sizes, but I’m on rural dial up and it’s S L O W !

Mike Russel:

In Photoshop, try using Lab mode, paying particu-
lar attention to the endpoints of the a and b
curves, to fix faded slides.

Yes. Thank you for the advince, Mike.

Sir F. A. Rien:

Quite, the whites are clipped and the blacks com-
pressed. in the horse image

How did you see they were clipped? The histogram is
not trimmed from either side…

The first Image is cropped so that you have no
film information to work with. You can crop
‘final’ later.

Yes, but I scanned an already cut slide in a special
holder. I didn’t have a full roll.

I scan to include 100% of the film chip. whatever
the ‘auto’ sets for the white and black, I expand
to include 10% more range, if possible. If the
result still shows compressed blacks or blocked
whites, re-do.

In some scanners if the blacks compress, you need
to -=reduce=- the exposure. It’s weird, but the
exposue setting is on the negative and then it’s
‘flipped’ to positive.

I usually set up my scanner exposure (total and per-
channel) so that each channel is "spread" over most of the dynamic range, becuase my scanner is only 12
bit per channel and this is more important than for
16-bit.

Also it is Ed Hammrick’s (author of VueScan) advice
to the set exposure by a transparent piece film
thereby making it as bright as possible (but not
overexposed). I think this is done to minimize the
noise-to-signal ratio.

I wonder why you say "compress" about the blacks (negative’s whites). If they are not overexposed and
there is no "clipping", isn’t it enough? Do sensors compress high-signal data?

Sorry for the smaller sizes, but I’m on rural dial
up and it’s S L O W !

Great thanks for doing them! I had already down-
scaled and downsampled the images before posting
anyway, but now I will have where to look back while
tinkering with curves.

Anton

Anton Shepelev <{oogle}mail.com> found these unused words:

Mike Russel:

In Photoshop, try using Lab mode, paying particu-
lar attention to the endpoints of the a and b
curves, to fix faded slides.

Yes. Thank you for the advince, Mike.

Sir F. A. Rien:

Quite, the whites are clipped and the blacks com-
pressed. in the horse image

How did you see they were clipped? The histogram is
not trimmed from either side…

The histogram doe snot show all the information. It ‘disappears’ at about 5-10% of the levels. I simply look at the image.

The first Image is cropped so that you have no
film information to work with. You can crop
‘final’ later.

Yes, but I scanned an already cut slide in a special
holder. I didn’t have a full roll.

There are mounts made for scanning that include the full frame and part of the biorder. In Weiss, IIRC, they’re "K". I prefer to scan the chp unmounted. That gives me the perforation holes as a reference for ‘pure’ light.

I scan to include 100% of the film chip. whatever
the ‘auto’ sets for the white and black, I expand
to include 10% more range, if possible. If the
result still shows compressed blacks or blocked
whites, re-do.

In some scanners if the blacks compress, you need
to -=reduce=- the exposure. It’s weird, but the
exposue setting is on the negative and then it’s
‘flipped’ to positive.

I usually set up my scanner exposure (total and per-
channel) so that each channel is "spread" over most of the dynamic range, becuase my scanner is only 12
bit per channel and this is more important than for
16-bit.

I’d not set individual channels at this point – just the overall exposure and I’d extend the automatic ‘white’ ‘black’ setting proposed.

Also it is Ed Hammrick’s (author of VueScan) advice
to the set exposure by a transparent piece film
thereby making it as bright as possible (but not
overexposed). I think this is done to minimize the
noise-to-signal ratio.

That’s why I scan the chip and use the perforations as ‘pure white’. It means the ‘white’ of the chip includes the slight reduction of the film. Actually you want to maximize the signal to noise ratio. The lower the ratio, the more noise.

You will get a bit more noise by capturing more of the ‘black’ and not accepting the histogram’s suggested low setting. You’ll also get more’noise’ in the whites as they’re not clipped. Mostly it’s the film support layer – it’s not ‘pure’.

I wonder why you say "compress" about the blacks (negative’s whites). If they are not overexposed and
there is no "clipping", isn’t it enough? Do sensors compress high-signal data?

When underexposed or when you set the histogram too high for the blacks, they are compressed. Let’s say from 0 to 16, is forced to 0. That then drags down the other levels and makes the dark areas ‘blocks’ rather than detail, even though dark.

You can stretch the darks, as with the horse, but it’s better to scan just outside the histogam’s shown ‘ends’ and then you can adjust with levels or curves. Once data is ‘blocked’, it’s gone.

Sorry for the smaller sizes, but I’m on rural dial
up and it’s S L O W !

Great thanks for doing them! I had already down-
scaled and downsampled the images before posting
anyway, but now I will have where to look back while
tinkering with curves.

Anton

Just curious, from your rememberance, how close was the coach?

JA

Sir F. A. Rien:

The histogram doe snot show all the information.
It ‘disappears’ at about 5-10% of the levels. I
simply look at the image.

Such is the graphical histogram, but you will not
find any pixel with a value of 255 in any channel.

There are mounts made for scanning that include
the full frame and part of the biorder. In Weiss,
IIRC, they’re "K". I prefer to scan the chp unmounted.

I have a Nikon Coolscan IV film scanner (and am
dreaming of Coolscan 5000) which has two adapters:

· An adapter for un-mounted 6-frame film strips.
It work most of the time, but if the film is
badly curled (either laterally or longitudi-
ally) it does not allow for sharp scans because
it can’t hold such film sufficiently flat. And
both ends of the strip are especially difficult
to get flat (and therefore sharp).

For important frames, I usually forcus at sev-
eral points on the film and then manually set
the focus to a value in-between the maximum and
minimum of the measures.

Even for the 2900 dpi resolution the effects of
little focus deviation (present even on seem-
ingly flat film) are apparent. The scanner’s
lens seems to have a narrow depth of field.

· An adapter for mounted 6-frame film strips. The
mounting itself is a "frame" that helps to hold film flat by clamping the strip along the edges
and along the inter-frame spaces. Being so fee-
ble as to yield somewhat to strongly curled
film’s desire to spring back into a roll, it is
nevertheless better for scanning deformed film.
And there’s lots of it in my father’s ar-
chieves.

It also can accept single frames in the "stan-
dard" mount, like the slides here demostrated.

I don’t know if it is possible to use another
adapter with this scanner. A glass film holder would
be interesting to try for it would hold any film
perfectly flat…

Sorry, I didn’t understand what Weiss is in your
context.

That gives me the perforation holes as a reference
for ‘pure’ light.

It is a nice idea to use perforation holes for that.
More so it is in slides where one is never sure what
the minimal density is, because it appears on maxi-
mally exposed areas, while in negatives you always
have transparent film as reference point. Thank you
for the advice.

I’d not set individual channels at this point –
just the overall exposure and I’d extend the auto-
matic ‘white’ ‘black’ setting proposed.

I hadn’t done it before I ran into a film so bad
that the range of one channel was two or more times
narrower that the that of the others, like the red
channel on the slides. Manually setting the indivi-
dual exposure in the way I mentioned helps to use
the sensor’s dynamic range more effectively and to
get a better (with more correct levels and the white
point, and with more data) raw scan requiring less
tweaking in the postprocessing application.

Actually you want to maximize the signal to noise
ratio. The lower the ratio, the more noise.

Yes, but I was mentioning the noise-to-signal ratio
🙂

When underexposed or when you set the histogram
too high for the blacks, they are compressed.
Let’s say from 0 to 16, is forced to 0. That then
drags down the other levels and makes the dark
areas ‘blocks’ rather than detail, even though
dark.

Hmmm. I wouldn’t call this compression. I’d call it
the clipping of blacks, or underexposure. If the
exposure is too low, densities from 0 to 16 are
"mapped" to, say, -20 to -4 on the censor that clips them all to zero. And with the clipping of whites
(overexposure) the situation is symmetric.

Compression is a non-linear operation, like a curve
to increase the contrast compresses blacks and
whites, while decompressing the mid-tones. The Lev-
els tool )(de)compresses the whole density range
uniformly. But neither happens with change of expo-
sure. That’s how I see it.

Just curious, from your rememberance, how close
was the coach?

I didn’t take this photos. They are older than I am
and originate from my friend’s collection. Also it
is not the original film (from the camera) but a
copy of it (which is probably the reason of the low
sharpness). In the Soviet times, such packs were
produced massively by Diafilm and sold at museums
e.t.c.

They also made "illustraded tales" based on anything from classics like "The Lost World" to Russian folk- lore. I have some of them too and am going to scan
them. They were in the half-frame format (24 x 18
mm), each frame consisting of a picture and several
lines of text.

Anton

Anton:

Hmmm. I wouldn’t call this compression.

Sorry for this. In the bus on my way home from work
it occured to me that I was wrong. Of course, com-
pression takes place too, because a higher exposure
setting causes the same density range on film to
occupy a wider brightness range in the digital scan.

Anton

Anton Shepelev <{oogle}mail.com> found these unused words:

Sir F. A. Rien:

The histogram doe snot show all the information.
It ‘disappears’ at about 5-10% of the levels. I
simply look at the image.

Such is the graphical histogram, but you will not
find any pixel with a value of 255 in any channel.

My scanner also has a histogram for helping set the exposure range. There I may see 228 as ‘suggested’ white. If I accept, that becomes 255 on the scan. I’ll set at about 242-248. Doing that I’ll see slight details in the whites that would otherwise be blocked [clipped].

There are mounts made for scanning that include
the full frame and part of the biorder. In Weiss,
IIRC, they’re "K". I prefer to scan the chp unmounted.

I have a Nikon Coolscan IV film scanner (and am
dreaming of Coolscan 5000) which has two adapters:

· An adapter for un-mounted 6-frame film strips.
It work most of the time, but if the film is
badly curled (either laterally or longitudi-
ally) it does not allow for sharp scans because
it can’t hold such film sufficiently flat. And
both ends of the strip are especially difficult
to get flat (and therefore sharp).

For important frames, I usually forcus at sev-
eral points on the film and then manually set
the focus to a value in-between the maximum and
minimum of the measures.

Even for the 2900 dpi resolution the effects of
little focus deviation (present even on seem-
ingly flat film) are apparent. The scanner’s
lens seems to have a narrow depth of field.

· An adapter for mounted 6-frame film strips. The
mounting itself is a "frame" that helps to hold film flat by clamping the strip along the edges
and along the inter-frame spaces. Being so fee-
ble as to yield somewhat to strongly curled
film’s desire to spring back into a roll, it is
nevertheless better for scanning deformed film.
And there’s lots of it in my father’s ar-
chieves.

It also can accept single frames in the "stan-
dard" mount, like the slides here demostrated.

I use an Epson Photo flatbed with illumination inf the lid. Current models are up to 4800 PPI! Not nearly as costly as the Nikon [you do have it so that’s a counter] and I can lay out a whole mess of chips, frame each, histgram each and scan as a batch.

Sorry, I didn’t understand what Weiss is in your
context.

weiss is the professional mounting firm. Perhaps not a prevalent now that multi-image shows aren’t being done with slide projectors, but created and shown as ‘video’. Their mounts were sprocket registered and they also made pin registration VR mounts.

That gives me the perforation holes as a reference
for ‘pure’ light.

It is a nice idea to use perforation holes for that.
More so it is in slides where one is never sure what
the minimal density is, because it appears on maxi-
mally exposed areas, while in negatives you always
have transparent film as reference point. Thank you
for the advice.

I think you’ll find that there is some opacity to even negative film. I’d still used the sprocked, though here’s it would set the BLACK level.

I’d not set individual channels at this point –
just the overall exposure and I’d extend the auto-
matic ‘white’ ‘black’ setting proposed.

I hadn’t done it before I ran into a film so bad
that the range of one channel was two or more times
narrower that the that of the others, like the red
channel on the slides. Manually setting the indivi-
dual exposure in the way I mentioned helps to use
the sensor’s dynamic range more effectively and to
get a better (with more correct levels and the white
point, and with more data) raw scan requiring less
tweaking in the postprocessing application.

Haven’t seen any wuite that bad. even the 1950’s Ansco Colour [nearly all ‘red’ now] responds to the ‘white’ eyedropper. I first set the pure white, note the value, then find a ‘white’ in the image and set the overall for that. Then manually set the overall RGB ‘white’ back to the previous number.

Actually you want to maximize the signal to noise
ratio. The lower the ratio, the more noise.

Yes, but I was mentioning the noise-to-signal ratio
🙂

When underexposed or when you set the histogram
too high for the blacks, they are compressed.
Let’s say from 0 to 16, is forced to 0. That then
drags down the other levels and makes the dark
areas ‘blocks’ rather than detail, even though
dark.

Hmmm. I wouldn’t call this compression. I’d call it
the clipping of blacks, or underexposure. If the
exposure is too low, densities from 0 to 16 are
"mapped" to, say, -20 to -4 on the censor that clips them all to zero. And with the clipping of whites
(overexposure) the situation is symmetric.

Both of which are undesirable.

Compression is a non-linear operation, like a curve
to increase the contrast compresses blacks and
whites, while decompressing the mid-tones. The Lev-
els tool )(de)compresses the whole density range
uniformly. But neither happens with change of expo-
sure. That’s how I see it.

This is linear compression with a ‘knee’, where in the rest of the image is shifted towards black, though retaining its linearity. I’m a video engineer, so we sued the terms compression and clipping a bit differently.

Just curious, from your rememberance, how close
was the coach?

I didn’t take this photos. They are older than I am
and originate from my friend’s collection. Also it
is not the original film (from the camera) but a
copy of it (which is probably the reason of the low
sharpness). In the Soviet times, such packs were
produced massively by Diafilm and sold at museums
e.t.c.

Oh yeah, have some 1950’s and 60’s ‘tourist packs’ – usually slammed through on the cheapest stock possible! My ‘available light’ shots in Carlsbad Caverns are still crisp, the folio bought there is junk now. Dyes faded, images grainy, etc.

They also made "illustraded tales" based on anything from classics like "The Lost World" to Russian folk- lore. I have some of them too and am going to scan
them. They were in the half-frame format (24 x 18
mm), each frame consisting of a picture and several
lines of text.

Anton

Have fun!

JA

Sir F. A. Rien:

Have fun!

Thank you very much, kind sir!

http://img411.imageshack.us/img411/6350/11exp.jpg

So old man Kokovanya, orphan Daryonka and cat Mury-
onka began to live together. They lived without
haste and were not avaricious and din’t complain
about their life for everybody had his own work.

Anton

Anton Shepelev <{oogle]mail.com> found these unused words:

Sir F. A. Rien:

Have fun!

Thank you very much, kind sir!

http://img411.imageshack.us/img411/6350/11exp.jpg

So old man Kokovanya, orphan Daryonka and cat Mury-
onka began to live together. They lived without
haste and were not avaricious and din’t complain
about their life for everybody had his own work.

Anton

But ‘who’ is the ‘cat’ in the window? <G>

lephoto.ttrr.org/animals/7601_021.html