HowTo? reduce 10000+ px panos to 1024 px?

[l_d_allan]

9.1.6 Pro

 

I'm getting marginal-to-poor results when reducing panos from 10,000 to 20,000+ px wide to 1024 px wide.

• I've tried setting the "Width" value in the "Create Panorama" tab to 1024, but that results in fairly obvious stair-stepping.
• Is that inherent in such a major down-res?
• My impression is that the setting for "Interpolation" is related to this.
• If so, which choice with 9.1.6 tends to be the "best practice"?
• Are there other settings that apply?
•  
• Are there workflow adjustments within PtGui that would result in less stair-stepping (and other artifacts)?
• Workflow adjustments outside of PtGui?
• Perhaps start with lower-resolution input instead of full-res 5600x3700?
•  
• I've attempted this resolution reduction with CS5 by itself after the pano was created, and
•  .... yuck.
• IIRC, in the past I tried multiple passes of CS5 as suggested by an article I came across, but
•  ... yuck.

 

 

[Ken Warner]

The stair step will happen with such a drastic resizing.

If I was going to do that, I use a good image editor like Photoshop or GIMP and make a series of half size source images until I got down close to what I wanted then use those to make the pano.

Each half size set should be smoothed and sharpened before resizing smaller. That's one fair to average way. There's no doubt some people here that have a much better workflow.

On 2/18/2013 8:06 PM, l_d_allan wrote:
> 9.1.6 Pro
>
> I'm getting marginal-to-poor results when reducing panos from 10,000 to
> 20,000+ px wide to 1024 px wide.
>

> - I've tried setting the "Width" value in the "Create Panorama" tab to

> 1024, but that results in fairly obvious stair-stepping.

> - Is that inherent in such a major down-res?
> - My impression is that the setting for "Interpolation" is related to
> this.
> - If so, which choice with 9.1.6 tends to be the "best practice"?
> - Are there other settings that apply?
> -
> - Are there workflow adjustments within PtGui that would result in less
> stair-stepping (and other artifacts)?
> - Workflow adjustments outside of PtGui?
> - Perhaps start with lower-resolution input instead of full-res
> 5600x3700?
> -
> - I've attempted this resolution reduction with CS5 by itself after the
> pano was created, and
> - .... yuck.
> - IIRC, in the past I tried multiple passes of CS5 as suggested by an

> article I came across, but

> - ... yuck.
>
>
>
>

[John Houghton]

On Feb 19, 4:06 am, l_d_allan <ldasig...@gmail.com> wrote:
>    - I've tried setting the "Width" value in the "Create Panorama" tab to

>    1024, but that results in fairly obvious stair-stepping.

>       - Is that inherent in such a major down-res?
>    - My impression is that the setting for "Interpolation" is related to
>    this.

Choice of interpolator is certainly important. The interpolators
supplied in the original Panorama Tools package were very prone to
show these aliasing artifacts when downsizing. Several new anti-
aliasing interpolators were later added. See:

http://wiki.panotools.org/Anti-aliasing_interpolators

This comparison shows how dramatic the difference can be between
interplators:

http://www.johnhpanos.com/interps.jpg

John

[PTGui Support]

But that applies only to the old panorama tools plugin. The
interpolators in PTGui are properly antialiased perfectly safe for
downsampling. There will be a difference between interpolators but it's
subtle.

John was interps.jpg created using panorama tools or PTGui?

Lynn since you consider the results from both PTGui and from CS5 to be
'yuck', I think you may have a wrong expectation here. Could you post
your result?

Kind regards,

New House Internet Services BV
Joost Nieuwenhuijse

-----------------------------------------------
PTGui - Photo Stitching Software

www.ptgui.com
For support see: http://www.ptgui.com/faq/
-----------------------------------------------

[PTGui Support]

On 19/02/2013 05:28, Ken Warner wrote:
> The stair step will happen with such a drastic resizing.

Before this starts to lead its own life: this is nonsense. Stair
stepping will not happen if proper antialiased interpolators are used.
The interpolators in PTGui are antialiased.

There may still be some issues in 'difficult' areas such as the nadir or
zenith of an equirectangular projection, but otherwise PTGui should give
smooth output. If not then this would be a bug.

Of course, when you downsample an image and then view it full screen,
you may see big blocky pixels. But that's not stair-stepping..

> If I was going to do that, I use a good image editor like Photoshop
> or GIMP and make a series of half size source images until I got down
> close to what I wanted then use those to make the pano.

This may have been the advice 20 years ago but with a proper
interpolator the result of downsizing in a single step should be just as
good. Just try it in Photoshop.

Joost

[John Houghton]

On Feb 19, 8:23 am, PTGui Support <supp...@ptgui.com> wrote:
> John was interps.jpg created using panorama tools or PTGui?

Joost, Those two compaison shots were taken from a set created in 2008
for several of the interpolators listed in the wiki article. As
PTGui's stitcher did not offer all of those options, they would have
been generated with a Panorama Tools stitcher - probably nona.

John

[PTGui Support]

The picture illustrates quite well the difference between an antialiased
and a non-antialiased interpolator. A non-antialiased interpolator
effectively becomes a nearest-neighbour interpolator when you are
downsampling by a large amount, and that's what you see happening in the
leftmost image. The rightmost image is properly antialiased.

Joost

[Gill747]

[Gill747]

I have often wondered what the interpolator setting in PTGui was for?

Is there any documentation that explains this setting and the other settings in the "advanced" create panoram ?

On Monday, February 18, 2013 11:06:26 PM UTC-5, l_d_allan wrote:

[Henrik]

Gill,

i don't have PTGui open, but from memory, you can specify the size in the last tap in the advance mode! you will obviously need to keep it in the proportion - i am not sure if there is a way you can set it so it crop the image to a say 1:10, but you can certainly set one side to 1024pixels

all the best from the hot and humid, North western Australia
 
Henrik Tived

----- Original Message -----

From:

pt...@googlegroups.com

To:

<pt...@googlegroups.com>

Cc:

Sent:

Tue, 19 Feb 2013 06:35:49 -0800 (PST)

Subject:

[PTGui] Re: HowTo? reduce 10000+ px panos to 1024 px?

--
--
You received this message because you are subscribed to the Google Groups "PTGui" group.
To post to this group, send email to pt...@googlegroups.com
To unsubscribe from this group, send email to ptgui+un...@googlegroups.com
Please do not add attachments to your posts; instead upload your files at a file sharing site (for example http://ge.tt/ ) and include a link in your message.
For more options, visit this group at http://groups.google.com/group/ptgui
 
---
You received this message because you are subscribed to the Google Groups "PTGui Support" group.
To unsubscribe from this group and stop receiving emails from it, send an email to ptgui+un...@googlegroups.com.
For more options, visit https://groups.google.com/groups/opt_out.
 
 

[PTGui Support]

Have you tried Help -> Help For This Screen ?

Joost

On 19/02/2013 15:35, Gill747 wrote:
> I have often wondered what the interpolator setting in PTGui was for?
> Is there any documentation that explains this setting and the other

> settings in the "advanced" _create panoram ?_

>
>
>
> On Monday, February 18, 2013 11:06:26 PM UTC-5, l_d_allan wrote:
>
> 9.1.6 Pro
> I'm getting marginal-to-poor results when reducing panos from 10,000
> to 20,000+ px wide to 1024 px wide.
>

> * I've tried setting the "Width" value in the "Create Panorama"

> tab to 1024, but that results in fairly obvious stair-stepping.

> o Is that inherent in such a major down-res?
> * My impression is that the setting for "Interpolation" is related
> to this.
> o If so, which choice with 9.1.6 tends to be the "best practice"?
> * Are there other settings that apply?
> *
> * Are there workflow adjustments within PtGui that would result in

> less stair-stepping (and other artifacts)?

> * Workflow adjustments outside of PtGui?
> o Perhaps start with lower-resolution input instead of
> full-res 5600x3700?
> *
> * I've attempted this resolution reduction with CS5 by itself

> after the pano was created, and

> o .... yuck.
> * IIRC, in the past I tried multiple passes of CS5 as suggested by

> an article I came across, but

> o ... yuck.

[Ken Warner]

Joost,

I'm sure you are right. The reason I suggested making intermediate versions of the source images was to allow tweaking of the image with other manipulations like bluring and sharpening to get just the image qualities one wanted. And to avoid the problems with "difficult areass".

I certainly don't want to start a lengthy discussion about the issue or try to convince anybody else that what I suggested was right or better. It's just one way of doing it. I will say no more....

Ken

[l_d_allan]

On Tuesday, February 19, 2013 1:23:02 AM UTC-7, PTGui Support wrote in response to OP from Lynn Allan:

 

• I've attempted this resolution reduction with CS5 by itself after the pano was created, and ...  yuck.

• IIRC, in the past I tried multiple passes of CS5 as suggested by an article I came across, but ... yuck.

 

• Lynn since you consider the results from both PTGui and from CS5 to be
  

• 'yuck', I think you may have a wrong expectation here. Could you post
  your result?
  

 

Sorry, I apparently wasn't clear. The results from both approaches using CS5 were "yuck". The PtGui result was much better, but still seems like it might have room for improvement. Before I started sliding sliders, spinning knobs, selecting radio-buttions, etc. in a trial-and-error approach, I thought I'd check with the forum experts whether there is a "best practice" (hopefully/preferably without a plug-in).

 

Lazy? Perhaps, but even if I was to find better settings or an improved workflow, I'd have only nominal confidence it was "best practice".

 

I was sure that someone on the forum has submitted a 20,000+ pano to some venue (club meeting with projector, contest, etc.) and figured out how to get better results than I've gotten.

 

And as my German heritage father would say to me when I was struggling and whining about something ... don't be the  "poor craftsman who blames his tools"

[Chris Erskine]

why do you need to downsize?

can you just use a player that supports multi-res pano?

chris

--

[l_d_allan]

On Wednesday, February 20, 2013 5:59:21 PM UTC-7, stupid....@gmail.com wrote:

why do you need to downsize?  

can you just use a player that supports multi-res pano?

 

It's become an issue in several circumstances, none of which have anything to do with virtual tours.

• http://www.meetup.com/Collaboration-of-Photographers/events/104167752/
• • the "ground rules" are limit of 1024 on the long side, to be displayed by projector
  • in the past, I’ve taken prints to the meetup-group because the down-sized files come out so poorly when using CS4 or CS5
  • doesn’t work all that well with prints in darkened meeting (understatement)
  • point of meeting is C&C (critique and comment)
  • scheduled for this coming Monday evening (Feb 25th)
  •  
• another local meetup-like group has monthly "P.O.M" (photo of month)
• • http://www.cosphotographers.com/  (currently down ... for maintenance?)
  • picture taken the previous month are uploaded ... typically 30 to 50 per month ... one per individual
  • critique and comment opportunities with pro's and semi-pro's in the group give feedback on the image submitted
  • “ground rule” is 800 px on longest side
  • berean.zenfolio.com/p813934691/h5603b20e#h5603b20e
  • ~ 16320 x 5204 original resolution pano-hdr with full moon on city and front-range
  • left as ProPhoto and compression=10 20Mb .jpg to compare to down-sized
  • also included HDR .tif files zoomed/cropped to ~ 200% to show the detail in the originals
  • can clearly make out "McCandless Truck Center" several miles away
  • http://berean.zenfolio.com/p813934691/h56043238#h56043238 underexposed for highlights
  • http://berean.zenfolio.com/p813934691/h56043238#h56043232 middle exposure
  • http://berean.zenfolio.com/p813934691/h56043238#h56043234 overexposed for shadows, front-range
  • http://berean.zenfolio.com/p813934691/h56047e08#h56047e08 arrow points to enlarged area
  • feedback C&C was something like “it looks way out of focus ... not sure why you submitted it”
    
  • Same master PanoHDR file but prep’ed for photo-sharing (sRGB and 10,000x3189 and compressed ~ 5 to 7)
    http://berean.zenfolio.com/p813934691/h3eff48e6#h3eff48e6
    
  • Same master, but down-sized to 800 x 255
  • http://berean.zenfolio.com/p813934691/h3eff48e6#h5603b202
  • note: the photo-sharing site may do some prep, so perhaps appropriate to download "original"
    
  • FYI: This is from the same overlook during another morning.
  • The Spanish Peaks to the south near the New Mexico border are 110+ miles away "as the crow flies"
  • were quite visible to the eye, but not to the camera sensor
  • downsized to 5120x1515 px from 30000x8877 px
  • http://berean.zenfolio.com/p813934691/h56047e08#h50db6d90
  • same area @ ~ 150% zoomed-in
    http://berean.zenfolio.com/p813934691/h5604f41a#h5604f41a
    
    
  • Note: all files have default ACR sharpening and NR
  • sorry for the long post

[Erik Krause]

Am 21.02.2013 01:45, schrieb l_d_allan:
> Sorry, I apparently wasn't clear. The results from *both approaches
> using CS5* were "yuck".

It all depends on what you want to achieve. So please state what you
want this downsized panorama for, show your current result and explain
why it is "yuck".

--
Erik Krause
http://www.erik-krause.de

[l_d_allan]

On Thursday, February 21, 2013 12:28:24 PM UTC-7, Erik Krause wrote:

It all depends on what you want to achieve. So please state what you
want this downsized panorama for, show your current result and explain
why it is "yuck".

Good questions ... here is my expectation:

to get a down-sized image from a ultra-high-res pano that looks as good as a normal non-pano that is down-sized.

 

Suppose I begin with a regular non-pano image from my Canon DSLR. It starts as about 5600x3700. Further suppose for purposes of discussion,  I crop it to 4500x3000. I've decided to take this image to the meetup-group "Collaboration of Photographers" this Monday for C&C, so I down-size it to the required 1024 width. Invariably, it looks pretty good, at least as far as retaining sharpness and clarity.

 

However, with panos like the example in the previous post that started as 16320x5204, the down-sizing by CS5 has been unsatisfactory. I'll repeat the links to the "before" and "after".

http://berean.zenfolio.com/p813934691/h5603b20e#h5603b20e

http://berean.zenfolio.com/p813934691/h5603b20e#h56047e08

 

Obviously, I'm not expecting the down-sized image to retain the detail of the original. In the output from PtGui, I can zoom in and read the "McCandless Truck Center" sign. I don't expect to be able to do that with the down-sized image. I do expect to get feedback other than "why did you submit this for C&C ... it looks way out of focus". (they didn't say 'yuck' ... out of politeness I suppose).

 

 

[UtahBob]

On Thursday, February 21, 2013 4:17:09 PM UTC-5, l_d_allan wrote:

On Thursday, February 21, 2013 12:28:24 PM UTC-7, Erik Krause wrote:

It all depends on what you want to achieve. So please state what you
want this downsized panorama for, show your current result and explain
why it is "yuck".

Good questions ... here is my expectation:

 I'll repeat the links to the "before" and "after".

In Erik's defense, I'll say I'm missing that long post with all the links.  Went I saw the follow-up post, I was scratching my head as to why I missed the post in my emails but I didn't.  Apparently that long post with the links and attachments never made it to my in-box.  I had to hit the forum to see it.  I don't have a spam box but my email host I believe does purge virus etc. emails so don't know what happened. 

IMHO, I don't know how I would even approach commenting on this given that I'm looking at a screen and not a 1024 size file projected but I don't find it out of focus - I just can't resolve anything with my eyes given what the photo shows.

Bob

[l_d_allan]

On Thursday, February 21, 2013 4:16:24 PM UTC-7, UtahBob wrote:

I'll say I'm missing that long post with all the links.

 

Possible, this is related to Google Groups re-implementing "Attach a File". I think that capability used to be available, then "it went away", and now it seems to be back.

[Keith Martin]

On 21 Feb 2013, at 21:17, l_d_allan <ldas...@gmail.com> wrote:

with panos like the example in the previous post that started as 16320x5204, the down-sizing by CS5 has been unsatisfactory. I'll repeat the links to the "before" and "after".

http://berean.zenfolio.com/p813934691/h5603b20e#h5603b20e

http://berean.zenfolio.com/p813934691/h5603b20e#h56047e08

Is this not just a difference in the method of interpolation? Did you use Bicubic Sharper in Photoshop's Image Size dialog or the default Bicubic? Does it help it you apply some careful sharpening to the downsampled graphic?

k

[dbd]

On Thursday, February 21, 2013 1:17:09 PM UTC-8, l_d_allan wrote:

On Thursday, February 21, 2013 12:28:24 PM UTC-7, Erik Krause wrote:

It all depends on what you want to achieve. So please state what you
want this downsized panorama for, show your current result and explain
why it is "yuck".

Good questions ... here is my expectation:

to get a down-sized image from a ultra-high-res pano that looks as good as a normal non-pano that is down-sized.

 ...

I've looked at the images on your links. You are comparing apples and eggplants. When an image that is mostly black with a low density of very bright impulses is down-sampled by 20 and projected at the same size as the original, the dark background looks unchanged, but the bright points have grown in size by about 20 times 20 or 400. That makes the bright points look out of focus.

 I do expect to get feedback other than "why did you submit this for C&C ... it looks way out of focus". ...

If you want other feedback, don't subject dark background/bright point images to desampling ratios that make them look out of focus. Your expectations are broken. Your night HDR does not survive desampling unaffected. Low dynamic range, high marking density images like your daytime image don't highlight the desampling artifacts to the same extent.

Dale B. Dalrymple

 

[UtahBob]

On Thursday, February 21, 2013 3:02:07 AM UTC-5, l_d_allan wrote:

• • can clearly make out "McCandless Truck Center" several miles away

• • http://berean.zenfolio.com/p813934691/h5604f41a#h5604f41a
    
    

Lynn, issues aside, I enjoyed these images for the shear ability to virtually drive through town. Perhaps images such as this should be enjoyed in their full resolution glory.   By the way, the McCandless Truck Center is 1.5 miles out - at least now I know where a good shot of town can be acquired.

Bob

[l_d_allan]

On Thursday, February 21, 2013 10:40:27 PM UTC-7, UtahBob wrote:

issues aside, I enjoyed these images for the shear ability to virtually drive through town. Perhaps images such as this should be enjoyed in their full resolution glory.   

 

Thanks for your kind and encouraging words ... but that option of "full res glosy" isn't an option at Monday's C&C session at the meetup group get-together..

 

By the way, the McCandless Truck Center is 1.5 miles out - at least now I know where a good shot of town can be acquired.

 

 

I infer from your userid that you live in Utah, not C.Springs. Did you used to live here? Get back for visits?

 

It's a Great! Location! that not many people seem to know about. In some ways, better than the GrandView overlook in Palmer Park.

 

 

[l_d_allan]

On Thursday, February 21, 2013 7:57:11 PM UTC-7, dbd wrote:

 ...

I've looked at the images on your links. You are comparing apples and eggplants. When an image that is mostly black with a low density of very bright impulses is down-sampled by 20 and projected at the same size as the original, the dark background looks unchanged, but the bright points have grown in size by about 20 times 20 or 400. That makes the bright points look out of focus.

 Your expectations are broken. Your night HDR does not survive desampling unaffected. Low dynamic range, high marking density images like your daytime image don't highlight the desampling artifacts to the same extent.

 

Ah, that explains A LOT! Thanks!  I'll either make prints for the C&C meetup, or capture+make+use images like you suggest. 

 

Bummer, because night-time panos are something I enjoy doing and are "outside the norm" of what people bring to the C&C sessions at our meetup club get-together. I'll try the down-sizing with CS5 and then PtGui on some very-high-res 360 degree interiors and see how they come out.

http://berean.zenfolio.com/zf_slides/h54be747e#h54bea1f6

http://berean.zenfolio.com/zf_slides/h54be747e#h54be9604

 

I had thought that the out-of-focus looking results might be due to jpeg using "block compression", but I recall getting similar results using .tiff's throughout the workflow.

 

To my shame, I'll acknowlege having plenty of low dynamic range images. :-)

I'm unclear what you mean by "high marking density images".

(Note that having the down-sized image shown by a projector doesn't really apply to the problem I've been grappling with. It's the .jpg itself from the 16000 px to 800 px down-size that appears out of focus).

 

[dbd]

On Friday, February 22, 2013 7:12:06 AM UTC-8, l_d_allan wrote:

...

To my shame, I'll acknowlege having plenty of low dynamic range images. :-)

In this case, low dynamic range is the characteristic of the daylight scene, not a reflection on the quality of the recording process.

I'm unclear what you mean by "high marking density images".

...

In the night-lights image most of the image is nearly black with a scattering of very bright pixels -> low marking density. The daylight image has very few nearly black pixels and almost all pixels brightly lit -> high marking density. Compare histograms and see how much more of the daylight image is to the right.

Dale B. Dalrymple

[l_d_allan]

In the night-lights image most of the image is nearly black with a scattering of very bright pixels -> low marking density. The daylight image has very few nearly black pixels and almost all pixels brightly lit -> high marking density. Compare histograms and see how much more of the daylight image is to the right.

Also Very Helpful! Again, thanks!

 

I'll rethink trying to down-size a low marking density image like this 360 degree star-trail pano :-)

http://berean.zenfolio.com/zf_slides/h54bea6d4#h54bea6d4

 

 

Dale B. Dalrymple

IIRC, a teacher (or principal?) at a grade school I attended in rural Missouri in the 50's was Mr. Dalrymple. Also a very nice gentleman. Any relation, by chance? 

[Erik Krause]

Am 21.02.2013 22:17, schrieb l_d_allan:
> Suppose I begin with a regular non-pano image from my Canon DSLR. It starts
> as about 5600x3700. Further suppose for purposes of discussion, I crop it
> to 4500x3000. I've decided to take this image to the meetup-group
> "Collaboration of Photographers" this Monday for C&C, so I down-size it to
> the required 1024 width. Invariably, it looks pretty good, at least as far
> as retaining sharpness and clarity.

I tried four different ways to downsample the image. I downloaded the
large version you supplied and converted it to 16 bit first. This was
the base for all tests. First I converted to Lab color space but this
gave almost the same result as RGB.

Next I tried Step interpolation, which means to downsample to 90% as
often as you need to get the final size. This gave a very different
result which kept many of the very small highlights in the background
but still gave that blurry look.

Then I tried Nearest Neighbor to downsample. This is a non-aliasing
interpolator which gives very hard pixels. It looks sharp but very
unnatural.

Last I used PTGui with Lanczos16 interpolator which has a slightly
sharpening very large kernel. This one gave the best compromise in my
opinion. Distant lights are kept as single pixels and edges are not aliased.

However, this image is not suited for downsampling. At least not that much.

[l_d_allan]

Ah, this is pretty much Exactly the kind of expert assistance I was hoping for. Thanks for tackling my problem.

 

BTW, I suppose you mean the downloaded 16320 px as the "base" (8 bit jpg ... 20Mb)

 

Based on the almost certainly valid input from dbd, I've "seen the error of my ways" on down-sizing "low masking density" panos. It appears that "you can't get there from here." :-)  You (and dbd) have saved me hours of unproductive trial-and-error.

 

However, I've got Plenty of "higher masking density" ultra-wide panos (uwp's ... 180 to 360 degree FOV):

http://berean.zenfolio.com/uwp

 

Have you heard the phrase, "no good deed goes unpunished"? I would Really Appreciate your (Erik's ... or dbd's) continued expert assistance on this, but on higher masking density panos that might actually down-size ok.

 

And rather than me uploading my images with original .tifs or even original .dng's for you to download and experiment with, how about tackling one or more of your own images that you have on your local disk? I don't think I'd be the only one to benefit from a description of the workflow you determine as your "best practice".

 

BTW: I recall your home page featuring a very nice image of a bridge in NYC. Or am I thinking of someone else? Night panos are of high interest for me, and I would be That Much More Appreciative of a "best practice workflow" you might be able to develop (and document) to get this (or similar) image to 2000 - 3000 pixels.

 

 

 

[Erik Krause]

Am 22.02.2013 23:06, schrieb l_d_allan:
> BTW: I recall your home page featuring a very nice image of a bridge in
> NYC. Or am I thinking of someone else? Night panos are of high interest for
> me, and I would be That Much More Appreciative of a "best
> practice workflow" you might be able to develop (and document) to get this
> (or similar) image to 2000 - 3000 pixels.

You probably mean Joergen Geerds "luminous New York". Yes, those are
exceptional. He intentionally shoots HDR and doesn't saturate the images
that much hence they are easier to downsize.

I just use photoshop bicubic sharper to downsample.

[l_d_allan]

Yes and Agree and book-marked both of your websites, plus dbd's "High and Wide"

On Friday, February 22, 2013 4:13:08 PM UTC-7, Erik Krause wrote:

You probably mean Joergen Geerds "luminous New York".  

Yes, those are exceptional. He intentionally shoots HDR and doesn't saturate the images
that much hence they are easier to downsize.

I just use photoshop bicubic sharper to downsample.  

 

Getting back to your informed experiments ... What PtGui setting(s) do you think would be best for a "low masking density" image like in the OP?

 

And getting back to the original question I wished I'd asked ... What PtGui setting(s) would you think would be "best practice" for a high or medium masking density image? (realizing that the most accurate answer would be "it depends" :-)

[UtahBob]

On Friday, February 22, 2013 9:53:49 AM UTC-5, l_d_allan wrot 

I infer from your userid that you live in Utah, not C.Springs. Did you used to live here? Get back for visits?

 

It's a Great! Location! that not many people seem to know about. In some ways, better than the GrandView overlook in Palmer Park.

Lynn,  stuck in NJ (not that it's a bad place - we have a rather amusing governor) - my user id reflects where I want to be...  I get to AZ, UT, CO and the like every so often.  Pikes Peak is on the bucket list so I should get a shot at some overlooks in your area.  Google Earth is great for identifying the shooting location of most panoramas that have some recognizable features.   I will look at Palmer Park also.       Bob

 

 

[Joergen Geerds]

I am a bit of a loss to understand the issue lynn has, or even to replicate it.

I do generate my high rez panos (10k-150k px width) with ptgui, save as PSB, and then downsample (after editing) via bicubic sharper in photoshop, and the result is fast and simple and perfect, and would be even if I have to deal with 500 images in one batch. no extra voodoo necessary.

even if you render a (originally 10k) wide project in ptgui directly to 1k, it looks ok... a bit more fuzzy than downsizing in photoshop, but certainly no aliassing steps or such, soft but acceptable.

Erik: thx for your kind words about my work :-)

[Joergen Geerds]

after reading a bit more superficially over the thread, there seems to be a problem with halo-ing and over-sharpening, which can create sub par down-sampled imaged on the web. this has nothing to do with ptgui. one of the best practices I "preach" is leaving any sharpening out of your process, until the very very last step, when you have the product in front of you downsampled to the size you need. downsample, then sharpen (if necessary), never any sharpening in any steps earlier than that, even RAW files have their sharpening dialed down to 0 in ACR. (this falls into my mantra of "never hurt original pixels")

[l_d_allan]

Joergen et'all,

If you had a chance to read the reply by dbd regarding "low masking density images" ...

 

> don't subject dark background/bright point images

> [[a.k.a. low masking density images]]

> to [[very high]] desampling ratios that make them look out of focus.

> Your expectations are broken.

> Your [[Lynn's]] night HDR does not survive desampling unaffected.

Do you find yourself in agreement or not with dbd's input? It seems reasonable to this down-sampling ignoramus, but I haven't made the time yet to try out his suggestions.

[l_d_allan]

UtahBob,

I've spent some time in western New Jersey, and saw why the state nickname is "Garden State".

On Friday, February 22, 2013 8:07:58 PM UTC-7, UtahBob wrote:

Lynn, stuck in NJ (not that it's a bad place -

Google Earth is great for identifying the shooting location of most panoramas that have some recognizable features. I will look at Palmer Park also.

Interesting. 

 

Well, if you ever get to the C. Springs area, let me know. I enjoy giving free guided tours to photographers. As the powder skiers say, “the locals know where the ‘secret stashes’ are,”   Can’t help you with where the O2 is hidden, however.

  

[l_d_allan]

Would you also "dial down to zero" the ACR/LR setting for "Clarity", which I believe is a form of sharpening of mid-tones?

 

On Sunday, February 24, 2013 12:57:05 PM UTC-7, Joergen Geerds wrote:

[Joergen Geerds]

On Monday, February 25, 2013 12:56:13 PM UTC-5, l_d_allan wrote:

Would you also "dial down to zero" the ACR/LR setting for "Clarity", which I believe is a form of sharpening of mid-tones?

absolutely, since clarity is very related to local contrast (the halo I mentioned) and it is a sharpening process (akin to USM with 50-200px radius and 20-50% power).