"Bill Pragnell" <bil### [at] hotmailcom> wrote:
> "Kenneth" <kdw### [at] gmailcom> wrote:
> > With a little bit of fog thrown in-- smog, or city pollution.
>
> This is a nice project, looking good. What happens if the fog is too dense to
> see the ground? :)

Answer: Too much fog! ;-) I spent hours trying to get a combination of regular
and groud fog that looked reasonably correct-- but it's still not right. (I may
do away with it altogether; some New York City aerial photos I've seen are
crystal clear, all the way to the horizon!)
>
> You could also try to group buildings in blocks, alley-distance apart, with
> blocks separated more widely. Maybe vary the height distribution using a
> large-scale pigment to select the maxima - you tend to get clusters of tall
> buildings in many cities.
>
I hadn't thought of that-- and you're absolutely right. For example, Manhattan
definitely has various-size cluters, mainly because of the depth of bedrock in
various parts of the city. Some areas just can't sustain the heavier loads, even
with pilings. That's a detail for my own city that I completely neglected-- and
it would add to the realism. Thanks!

Post a reply to this message

Am 8/5/2017 um 23:27 schrieb Kenneth:
> "Kenneth" <kdw### [at] gmailcom> wrote:
>> Another test-image post-- hopefully my final gamma test... ;-)
>>
>> This PNG render is post-processed in Photoshop, but I CHANGED my usual 2.2 gamma
>> there (in PS) to 1.0. *IF* this image shows up correctly here-- in both small
>> and large previews-- then I will at least have solved the major part of my *own*
>> problem...
>>

actually this PNG image has a embedded gamma of 2.0 and the image you 
did post earlier (the dark one) has a gamma chunk of 4.4 - and this 
doesn't make any sense at all.
>
> Success!! Both preview sizes here look identical, and correct--well, with maybe
> a *very slight* washed-out appearance in both-- but that could be from a number
> of reasons on my end, and is unimportant in the bigger scheme of things ;-)
>
> SO... it seems that my use of Photoshop-- at its *typical* gamma setting of
> 2.2-- was introducing a gamma change into the POV-Ray PNG image. I have aways
> assumed 2.2 to be the correct gamma in PS, for any/all PNG images that I've
> downloaded from the 'net, and even for Photoshop-created PNG images. But it's
> NOT correct for post-processing a POV-Ray PNG render: a PS gamma of 1.0 is
> necessary (at least for both of the newsgroup's two preview sizes to show up
> correctly!) So, *something* is amiss somewhere-- either a flaw in my version of
> Photoshop, OR in how the newsgroups treat preview images.
>

Which PS version do you use? Adobe changed the handling of various image 
file format between different versions a lot and especially Photoshop's 
handling of PNG in anything older than CS6 is known to be flawed.

I usually do recommend to open images in PS that do not include an ICC 
profile in a way that that PS does NOT assign a "working colorspace"
and also does NOT assign a color profile.

-Ive

Post a reply to this message

Ive <ive### [at] lilysoftorg> wrote:
> Am 8/5/2017 um 23:27 schrieb Kenneth:
> >
> > Another test-image post-- hopefully my final gamma test... ;-)

>
> actually this PNG image has a embedded gamma of 2.0 and the image you
> did post earlier (the dark one) has a gamma chunk of 4.4 - and this
> doesn't make any sense at all.

WOW, that's... crazy. I don't even know what to make of it. *Thanks* for
checking this out. One idea that occurs to me (which could be wrong, of course)
is that posting a PNG to the newsgroups changes the gamma... perhaps twice, for
the two different-sized previews. ?? Otherwise, it's a mysterious problem with
several interrelated causes, my Photoshop being only one of them.

If you don't mind doing so, would you check the gamma chunk in the 'raw' POV-Ray
PNG image I posted, the one that begins with this:
"One more image test, again for my own purposes: This PNG is *directly* from
POV-Ray (and NOT post-processed in Photoshop.)..."

I'm curious to know what *that* turns out to be.
>
> Which PS version do you use? Adobe changed the handling of various image
> file format between different versions a lot and especially Photoshop's
> handling of PNG in anything older than CS6 is known to be flawed.
>

Version 5.0, I'm embarassed to say. And yes, I also read somewhere that such
'older' versions had some kind of flaw in their handling of PNGs... although I
don't know the technicalities.

Post a reply to this message

"BayashiPascal" <Pas### [at] BayashiInJapannet> wrote:
> That's really nice.
> I think you should also make different photos for the 1st floor of the
> buildings. At least to give them an entrance.

YES, I've been thinking the same; it's on my to-do list now. A ground
floor/lobby would add a LOT to the look of the rather boring facades. I'm not
yet sure how I'm going to go about it, though-- with (more) photos, or maybe
with some kind of fancy procedural pigment instead, that has some randomness to
it.

> Do you plan to add people, vehicles and traffic signs in the street ? It looks
> like an endless project ! :-)

I'm thinking of adding some moving cars at least-- as soon as I get around to
making some ;-) Yeah, this city project has taken on a life of its own. But I
discover some new techniques every time I add or change something, so it
continues to be a nicely challenging coding exercise.

Post a reply to this message

> WOW, that's... crazy. I don't even know what to make of it. *Thanks* for
> checking this out. One idea that occurs to me (which could be wrong, of course)
> is that posting a PNG to the newsgroups changes the gamma... perhaps twice, for
> the two different-sized previews. ?? Otherwise, it's a mysterious problem with
> several interrelated causes, my Photoshop being only one of them.
>

There *is* a problem with the web interface as the code that creates the 
downsampled preview always assumes an input image with 2.2 (or sRGB) 
gamma. It takes the original image data, resamples it and does tag the 
output as sRGB regardless of the original gamma information.
On the other hand, how the image itself appears solely depends on the 
way the browser handles it. And luckily we live in 2017, 10 years (or 
so) ago literally EVERY browser did handle it differently.
And BTW the problem is not limited to PNG images, it would also happen 
when you use for instance a JPEG with an ICC profile stating a non 2.2 
gamma.


> If you don't mind doing so, would you check the gamma chunk in the 'raw' POV-Ray
> PNG image I posted, the one that begins with this:
> "One more image test, again for my own purposes: This PNG is *directly* from
> POV-Ray (and NOT post-processed in Photoshop.)..."
>
> I'm curious to know what *that* turns out to be.

As expected - this one has a sRGB chunk.
(Rendered with POV-Ray 3.7.1-beta.9 at 22:22, 2017-08-04)


> Version 5.0, I'm embarassed to say. And yes, I also read somewhere that such
> 'older' versions had some kind of flaw in their handling of PNGs... although I
> don't know the technicalities.
>

I've worked with all versions of Photoshop before they turned to CC and 
Adobe's handling of some image file formats was always a mess. But for 
PNG it was the worst of all as Adobe did - when switching versions -  do 
it wrong again, but in a different way. So e.g. a PNG file written with 
PS 5.0 will look unexpected different when opened in CS1.
Good news is, as a user you can work around these issues. But as I've 
used version 5.0 quite a long time ago (20 years ?) I cannot tell 
anymore how the settings are called and where to find them.
Look somewhere at Preferences->Settings->Color management. There should 
be somewhere a checkbox saying something like "always ask when profile 
differs" and make sure it is checked. When opening now e.g. a PNG file 
written by POV-Ray a dialog should pop up asking you if you would like 
to apply a color transform and you should say NO.
But as mentioned, I'm not entirely sure, maybe this was the way v6.0 did 
handle it.

-Ive

Post a reply to this message

"Kenneth" <kdw### [at] gmailcom> wrote:

Kenneth, this is coming along really well!

I think Bill Pragnell already mentioned adding a mechanism by which to cluster
groups of similar-sized buildings - if you use a color-coded map to do something
like that, you could detail things like high-rises, small residential houses,
shopping areas with large parking lots, etc - but you could also then add in
landscape features like trees, rivers, parks, etc.

And just to stretch the imagination - I wonder if you could use such a scheme to
vary the atmosphere.

I'm always interested in your approach to handling this - and I'me sure new
users would be interested in seeing some of the magic behind making something
like this work, and others would find it easier to solve problems / suggest
improvements if they had an idea of what your [pseudo]code is.


How large can you make this?
This almost begs one to take on the challenge of a "world-building" scene like a
video game such as minecraft, etc.
(I have no idea how so much gets generated and handled with just a tablet or
cell phone)
I could envision adding a maze-solving algorithm to guide the camera along the
streets, and doing a very long fly-through animation to just show that it can
keep going, and going, and going...

Along those lines, perhaps add in "under construction" as a building type.

User "Architype" has his wonderful historic-style architectures --- I could
envision an interesting scene where the farther along in the animation one goes,
the more modern the architecture gets   :O

Maybe think about doing an old / odd building or two to serve as theaters /
museums / concert halls, etc.

This is really coming along spectacularly :)  I'm very glad you've been able to
make so much excellent progress and work out so many things on your to-do list.

Thanks for sharing!

Post a reply to this message

On 2017-08-05 06:15 PM (-4), Kenneth wrote:
> "Bill Pragnell" <bil### [at] hotmailcom> wrote:
> Answer: Too much fog! ;-) I spent hours trying to get a combination of regular
> and groud fog that looked reasonably correct-- but it's still not right. (I may
> do away with it altogether; some New York City aerial photos I've seen are
> crystal clear, all the way to the horizon!)

Probably because they chose the photo shoot date very carefully.  I have 
never seen a clear horizon when I've visited in NYC; usually, I could 
not see more than 10 miles (15 km) or so.

>> "Kenneth" <kdw### [at] gmailcom> wrote:
>> You could also try to group buildings in blocks, alley-distance apart, with
>> blocks separated more widely. Maybe vary the height distribution using a
>> large-scale pigment to select the maxima - you tend to get clusters of tall
>> buildings in many cities.
>>
> I hadn't thought of that-- and you're absolutely right. For example, Manhattan
> definitely has various-size cluters, mainly because of the depth of bedrock in
> various parts of the city. Some areas just can't sustain the heavier loads, even
> with pilings. That's a detail for my own city that I completely neglected-- and
> it would add to the realism. Thanks!

IIRC, the NYC code requires its skyscrapers to have some kind of recess 
or step architecture to minimize the wind effects at ground level.  This 
limits how close the actual towers can be to each other.

Post a reply to this message

> "BayashiPascal" <Pas### [at] BayashiInJapannet> wrote:
>> That's really nice.
>> I think you should also make different photos for the 1st floor of the
>> buildings. At least to give them an entrance.
> 
> YES, I've been thinking the same; it's on my to-do list now. A ground
> floor/lobby would add a LOT to the look of the rather boring facades. I'm not
> yet sure how I'm going to go about it, though-- with (more) photos, or maybe
> with some kind of fancy procedural pigment instead, that has some randomness to
> it.

Maybe make the ground floor very slightly larger than the rest.


dimentions.

Post a reply to this message

"Bald Eagle" <cre### [at] netscapenet> wrote:
> "Kenneth" <kdw### [at] gmailcom> wrote:
>
> Kenneth, this is coming along really well!

Thanks! Every new change that I make leads to another idea that I want to try.
>
> I think Bill Pragnell already mentioned adding a mechanism by which to cluster
> groups of similar-sized buildings - if you use a color-coded map to do
> something like that, you could detail things like high-rises, small
> residential houses, shopping areas with large parking lots, etc - but you
> could also then add in landscape features like trees, rivers, parks, etc.

Hmm, that could take awhile. I wonder how many years I have left to live! :-P I
do LIKE the ideas, though. Not sure about the residental houses and shopping
malls, though; older megalopolis cities (like Manhattan) have such expensive
real-estate that almost everything needs to be built vertically, with
multi-storeys, and crammed together. (I wonder if there is a *one*-storey
building to be found anywhere on that island!)
>
>
> I'm always interested in your approach to handling this - and I'me sure new
> users would be interested in seeing some of the magic behind making something
> like this work, and others would find it easier to solve problems / suggest
> improvements if they had an idea of what your [pseudo]code is.

I originally thought I could put the basic 'single building' SDL code into a
macro, to be duplicated (and easily explained!) Perhaps that's still possible,
but the scene has gotten more and more complex (along with what seems like a
limitation in how POV-Ray's image_pattern works-- a very important part of the
code that I can't seem to simplify.) So right now, it's more like an organic
living 'thing'-- that even *I* don't fully understand, ha. Later on, I *might*
be able to explain the approach I used in a simplified and abstract way; but I'm
still working it all out.

>
> How large can you make this?

Up to the limit of a computer's RAM and hard drive! ;-) I only need to change
one number-- from the current 450 (buildings) up to, say, 10,000, etc. Of
course, the *parse* time is going to increase mightily.

By the way, POV-Ray's 3.7xx 'clockless animation/real-time-raytracing'
experimental feature has come in REALLY handy, for doing quick camera-animation
flyovers of this city scene. It saves SO much extra parsing time. Highly
recommended ;-) (I hope the developers find some way to SAVE the animation
renders as actual image files; currently the fast animation just renders
on-screen.)

> This almost begs one to take on the challenge of a "world-building" scene
> like a video game such as minecraft, etc.
> (I have no idea how so much gets generated and handled with just a tablet or
> cell phone)...

Although I don't know much at all about graphics processing units (GPUs), I
think such seemingly 'infinite' worlds are generated in there, for the most
part. (I'm out of my league with this topic!)
>
> Along those lines, perhaps add in "under construction" as a building type.
>
I like it!!  Didn't think of that. Then it will give me a chance to build some
tall construction cranes too.
>
> Maybe think about doing an old / odd building or two to serve as theaters /
> museums / concert halls, etc.

Yes indeed, if time permits (well, if my patience holds out, ha); the city
definitely needs some 'variety'-- at least some building shapes that aren't just
glorified boxes.

And I'm *still* considering replacing the box shapes with heightfield slabs.

Post a reply to this message

Ive <ive### [at] lilysoftorg> wrote:

>
> There *is* a problem with the web interface as the code that creates the
> downsampled preview always assumes an input image with 2.2 (or sRGB)
> gamma. It takes the original image data, resamples it and does tag the
> output as sRGB regardless of the original gamma information.

Aha! So I wasn't SO crazy after all ;-)  I much-appreciate your good detective
work. I need to digest all of this, to make some sense of it re: using my older
version of Photoshop. (*IF* I could find some newer standalone CS version of
PS-- for sale outright, and not 'rented' monthly from Adobe-- I would buy it,
and chuck my old version. It has served me well, but definitely needs an
update.)

> On the other hand, how the image itself appears solely depends on the
> way the browser handles it. And luckily we live in 2017, 10 years (or
> so) ago literally EVERY browser did handle it differently.
> And BTW the problem is not limited to PNG images, it would also happen
> when you use for instance a JPEG with an ICC profile stating a non 2.2
> gamma.

Interesting! (By the way, I'm running Windows 7 Ultimate, with the latest
edition of Firefox.)

>
> ...a user you can work around these issues. But as I've
> used version 5.0 quite a long time ago (20 years ?) I cannot tell
> anymore how the settings are called and where to find them.

I've included a screenshot of the only pertinent dialogs I can find in PS 5.0.
(Plus my computer's video-chip info.) And I think I discovered an erroneous
setting in the 2nd panel ("Profile To Profile"): RGB to LAB? That doesn't look
right. I'm guessing it should be SRGB to SRGB(?) This level of technical detail
is mysterious to me, honestly; I need to read up on those dialog boxes, in my
handy "Photoshop 5.0 Bible."

Thanks again for your help!

Post a reply to this message

Attachments:
Download 'ps_5_misc.jpg' (159 KB)

Preview of image 'ps_5_misc.jpg'