Banding

[klausesser]

Re: Banding

by klausesser » Thu Feb 20, 2014 2:19 pm

Well - and what´s that supposed to tell us in terms of the issue we talk about here?

best, Klaus

It tells us that PTGui is no longer simply a GUI-front end to Panorama Tools and has not been for many years now. It now incorporates ideas from many other sources.

Ok - i choose the wrong words, sory: i didn´t mean that PTGui today still is "just" a GUI for the PanoTools.

P.S.: you see - when PTGui can deal perfectly wit the lens without banding - why would APG fail?

Because PTGui incorporates vignetting correction and APG does not (yet); it also has different colour correction and blending systems/'engines'.

What about the lens correction models in the editor??

Klaus

Last edited by klausesser on Thu Feb 20, 2014 2:27 pm, edited 2 times in total.

[klausesser]

Re: Banding

by klausesser » Thu Feb 20, 2014 2:24 pm

Hallo Klaus
Ich habe hier keine Lust, mich zu streiten, was bei mir passiert oder nicht.

Es geht ja nicht ums Streiten, Martin. Es geht darum, Ursachen und Wirkung zu differenzieren.

Das geht nur wenn wir uns bemühen, sie faktisch auseinderzuhalten . .

Gruß, Klaus

[Boley]

Re: Banding

by Boley » Thu Feb 20, 2014 9:36 pm

Geben wir hier doch Robert erst mal Gelegenheit, die vielen Anregungen auszutesten und schauen dann, was zum Ziel führt, ja ? Sonst verliert er auch noch den Überblick.

There are many options to try and sometimes the only way is to eliminate or confirm suspicions. Do you think that either more or less overlapping may help in this instance?

btw.: are you living in Australia? Did we get in contact via Josef?

Hi Klaus, it is the same Robert and I have been in contact with you through Josef. Josef is still working on the problem which appears to be very baffling.

Thank you all for sharing your knowledge and experience, you are the Gurus of stitching and are well admired. These forums are great where the manufacturers of the product does not provide, in my opinion enough documentation and explanations for such a great and powerful software package.

I will let you know how I go with my tests in the next week or so and hopefully I may have something to share which is of value to others.

best regards
Robert

[lumelix]

Re: Banding

by lumelix » Thu Feb 20, 2014 9:53 pm

Ja, Du hast recht. Ich versuche das auch immer. Ich habe Robert mal den Link zum Lens Profile Creator durchgegeben. Den braucht er sowieso, auch wenn er das Lens-Plugin in APG verwendet. Ich denke, wenn er die Vignettierung dieser Bilder sauber weg bekommt ist das Banding dann auch weg.

Der Vorteil des Plugins könnte sein, dass APG zumindest weiss, dass da schon was korrigiert wurde (wenn wir Glück haben). Ich bin auch kein Fan von vorgängigen Modifikationen. Gibt meistens dann Mist in APG. Aber wenn es nicht anders geht ...

Dieses Banding trat bei mir früher sporadisch mal auf bei GIGA-Aufnahmen, die ganz leichte Vignettierung (auch an der langen Seite) aufwiesen und rel. wenig Überlappung <20% hatten. Das sie nur als vertikale Bänder und nicht auch horizontal auftreten ist unlogisch, aber trotzdem wohl ein Fakt.
Das deutet darauf hin, dass APG die horizontalen Übergänge besser hinbekommt als die Vertikalen. Oder eben, dass das Blending bei den vertikalen Übergängen irgend einen Mist macht.

Das würde sich auch mit dem anderen Riesen-Problem decken, dass mich nun langsam tödlich frustriert: Das Blending erzeugt (mit zig Varianten an Ankern etc.) bei vielen Bildern (>30 horizontal) immer noch viele Fehler und diese Farb- und Helligkeitsverläufe von links nach rechts. Ich habe das vor über 2 Jahren zum ersten mal gepostet als wir noch Version 2 hatten. Damals wurde ich auf Version 2.5 vertröstet, dann auf Version 3 und dann auf 3.5.

Nun haben wir 3.5 Beta2 und das Problem besteht immer noch völlig unverändert !!! Aus meiner Sicht ist APG so nicht in der Lage, brauchbare GIGA-Panoramen zu erstellen. Denn es ist nicht möglich hunderte oder sogar tausende von Bildern mit exakt der gleichen Beleuchtung aufzunehmen. Das Blending ist essenziell, ohne geht es einfach nicht.
Aus meiner Sicht müsste das G aus APG gestrichen werden!

Das Stitching und die Passpunktdetection wurde klar verbessert in dieser Version. Leider führt das auch dazu, dass ich erst mal die Settings richtig einstellen musste, damit er mir alte Test-Panoramen von damals richtig zusammen setzt. Wie immer fehlt bei all den tollen neuen Features total das Manual, respektive sogar die Begriffserklärung. Z.B. bei der Optimierung gibt es jede Menge (nicht ins Deutsche übersetzte) neue Modelle und Methoden mit völlig unverständlichen Bezeichnungen. Ich hatte Glück und habe gleich die "richtigen Knöpfe" gedrückt, rein intuitiv! weil ich diese Kiste doch schon etwas kenne. Wenn du Pech hast, kannst Du da tagelang rumprobieren ohne Erfolg.

Nun frage ich mich halt wirklich, warum ich diese Software wohl hunderte von Stunden ausgetestet habe und immer wieder aufwändige Bug-Reports mit vielen Bildern und Hinweisen geliefert habe, wenn das Zeug halt immer noch nicht funktioniert. Bringt doch nix, wenn die immer mehr Features reinpacken, aber die essentiellen Dinge wie das Blending funktionieren überhaupt nicht.

Gruss aus Basel

Last edited by lumelix on Thu Feb 20, 2014 10:12 pm, edited 2 times in total.

[lumelix]

Re: Banding

by lumelix » Thu Feb 20, 2014 10:09 pm

There are many options to try and sometimes the only way is to eliminate or confirm suspicions. Do you think that either more or less overlapping may help in this instance?

Hi Robert
Sorry that I wrote this message to Klaus in German, but I can't bothered then to translate this.

How much overlapping did you have in your case?
I found, that a good value is 25%. If you have a lens with a big aperture (and also more vignetting) like this f1.4 85mm, 30% overlapping can result in better Panoramas. It's simple because you need less area from the Corners which have some vignetting.

If you can create a specially lens Profile for your lens, try out first the Lens-Profile-Plugin in APG. If this can't fix the issue, try out to modify the Images first in LR or PS and check then if the vignetting is gone. With this images you shouldn't get this banding again.

If you like, you can upload your Images (please, downsize it to 50% and use JPG Quality max. 70%) to a file sharing platform, so we can download it and try out what's going wrong. Most of time, we found a solution to get this APG to work like it should

[Boley]

Re: Banding

by Boley » Thu Feb 20, 2014 10:34 pm

Thanks Martin, I use 20% overlap' perhaps I will try 25 or even 30%, but of course that means more images. If it rectifies the problem then so be it.

One of the reasons that I photographed column by column rather than row by row as I normally do was to get consistency in the lighting as it was evening and light was slowly fading. In the past I would take shots by rows and by the time the pan head came around again there would be a noticible difference in light between two rows, especially in the sky. Outdoors you are faced with many challenges as opposed to indoors where everything can be controlled.

I will try the APG plugin, colour anchors and PTgui and which of these is the easiest to rectify the problem. As a last resort I will take up your offer and download the images into Dropbox for you to look at, but, I do not wish to waste your time as I am sure you are busy with you own projets and issues.

I agree with you when you say "what is the point of all these new features if they are unable to get critical and basic things right like banding correct" - makes you wonder.

Thanks again

Kind regards
Robert

[klausesser]

Re: Banding

by klausesser » Fri Feb 21, 2014 1:48 am

Das Stitching und die Passpunktdetection wurde klar verbessert in dieser Version. Leider führt das auch dazu, dass ich erst mal die Settings richtig einstellen musste, damit er mir alte Test-Panoramen von damals richtig zusammen setzt. Wie immer fehlt bei all den tollen neuen Features total das Manual, respektive sogar die Begriffserklärung. Z.B. bei der Optimierung gibt es jede Menge (nicht ins Deutsche übersetzte) neue Modelle und Methoden mit völlig unverständlichen Bezeichnungen. Ich hatte Glück und habe gleich die "richtigen Knöpfe" gedrückt, rein intuitiv! weil ich diese Kiste doch schon etwas kenne. Wenn du Pech hast, kannst Du da tagelang rumprobieren ohne Erfolg.

Nun frage ich mich halt wirklich, warum ich diese Software wohl hunderte von Stunden ausgetestet habe und immer wieder aufwändige Bug-Reports mit vielen Bildern und Hinweisen geliefert habe, wenn das Zeug halt immer noch nicht funktioniert. Bringt doch nix, wenn die immer mehr Features reinpacken, aber die essentiellen Dinge wie das Blending funktionieren überhaupt nicht.

Da kann ich Dir nur voll und ganz zustimmen!

On the other hand see the London Gigapixel - no banding . . so it must be possible.

Did you stick at 20% overlap or did you try 25% also? The Nikkor 1,4/85mm definitely has no vignetting @f8 and f11 at all which could cause the banding. Otherwise PTGui must show banding also.

IF the reason is the blending algo . . . then it´s useless for projects with hundreds of shots. But i cannot believe that.

Gruß aus Düsseldorf, Klaus

[klausesser]

Re: Banding

by klausesser » Fri Feb 21, 2014 2:00 am

I found, that a good value is 25%. If you have a lens with a big aperture (and also more vignetting) like this f1.4 85mm, 30% overlapping can result in better Panoramas. It's simple because you need less area from the Corners which have some vignetting.

Sorry for bumping in here - but i tested the 1,4/85mm once more: there´s definitely no vignetting at all @f8 or f11 which might cause banding. So i strongly suggest NOT to use pre-stitch processing of lens parameters. Let the stitcher do it. You already mentioned very truely that we lack of precise descriptions which lens-model works in which way exactly and which distortion-model should be used preferably for which task.

THIS would be vital informations.

best to you, Klaus

[klausesser]

Re: Banding

by klausesser » Fri Feb 21, 2014 2:04 am

Hi Klaus, it is the same Robert and I have been in contact with you through Josef. Josef is still working on the problem which appears to be very baffling.

Hey Robert!

Josef visited me today and we took measures from the D800 - all values are what they are supposed to be!

I´ll drop you a mail today!

best to you, Klaus

[lumelix]

Re: Banding

by lumelix » Fri Feb 21, 2014 6:23 am

... , yes there still appears to be some vignetting but quite minor, certainly better than without the correction.

Hi Klaus
Robert wrote in the fifth post, that he HAS vignetting. More without LR modification than with. This is why I have suggested, the banding is caused by vignetting.
I never saw a f1.4 lens totally without vignetting. Perhaps there are also differences between some lenses. Or he use a special sunshade that will cause the vignetting. I remember that I do so as I shot a GIGA with deep standing sun.

[klausesser]

Re: Banding

by klausesser » Fri Feb 21, 2014 3:57 pm

Hi Klaus
Robert wrote in the fifth post, that he HAS vignetting.
[/quote]

Hi Martin!

I read it - but i doubt it. I have the same lens - i tested it critically and get extremely low vignetting @f4 and NO recognizable vignetting at all @f8. That´s why i usually suggest to use f8.

I never saw a f1.4 lens totally without vignetting.

The Nikkor f1.4/85mm AIs? At which aperture?

Perhaps there are also differences between some lenses. Or he use a special sunshade that will cause the vignetting. I remember that I do so as I shot a GIGA with deep standing sun.

In Robert´s image we can clearly see that there was no deep standing sun . .

Even if he had used a wrong sunshade: the vignetting caused by it would occur on the short side of the frame - not on the long side as it is implicated by his image showing vertical stripes.

So my conclusion:
1) IF the issue really is with vignetting: i suspect it´s caused by pre-processing.
2) IF the issue is NOT caused by lens-vignetting maybe it´s sunshade-vignetting. But even then: it´s on the wrong side of the adjacent images´s frames.
3) IF the issue is caused by the blending algos: why do they occur in APG, but not in PTGui?
4) given it´s caused by the algos - which i dont believe - then the algos in APG would be hardly usable for Gigapixels compared to the algos in PTGui.
5) but there are some Gigapixels done with APG without banding . . .

So i guess the issue is caused by some settings.
The problem is: the tutorials/descriptions regarding lens-distortions and familiar items just scratches the surface - there is no vital informations what they do and in which case which setting is preferable.

HERE i assume the real issue.

best, Klaus

Vignetting of a Nikkor AIs 1,4/85mm:

[Boley]

Re: Banding

by Boley » Sun Feb 23, 2014 5:00 am

Thanks for doing the tests Klaus.

Since my last entry I did some work and I must say with great success.

First I tried the APG lens plugin and my system crashed, with both 2.0.8 and 3.5

I then then download the Adobe Lens Profiler softeware and proceded to make some profiles for for this lens. I created 3 profiles, f16, f11 and f8. As my landscapes require a lot of foreground I require depth of field and don't assume shooting any lower but in any case looking at Klaus's results there would be minimal vignetting below than f stop anyway.

I then loaded the profile into my LR profile folder and corrected the images. I exported to APG and the result was amazing. It improved the banding by 90% and then with some colour anchors the result is as below.

I set blending to 0, ISO and diamond (but I don't think that did anything).


When I did these corrections I still noticed banding, although not as bad as previous. I thought was a great improvement and could fix it in PS but after rendering all the banding was eliminated. I always thought the final rendered image is the same as the edited image in APG.

I forgot to mention I did get the APG lens correction plugin to work but it uses Adobe profiles anyhow so I am not sure if it applies it differently to LR or PS.

In regards to the lens shade, I am using the standard shade that comes with this lens.

If anyone is interested in these profiles I will be more than happy to send them to you, just let me know. They were created with my D800E for the Nikkor AI-S MF 85.1.4 lens - 3 separate profiles at f16, f11 and f8 as stated above.

I have since photographed another set of images this time shot at f11 and f8 but have not processed yet and will do tests with and without profiles and will let you know later in the week.

On another topic I have a query on colour anchors. They helped with this issue and I will tend to use them more in the future. I have a question on the yellow reference points. How do you delete or move them? I could only add them. Also, if you have more than one point which one are the images reference to when you do the corrections?

Thanks for all your help. If it wasn't for this forum I would be stuffed, the Kolor documentation is less than adequate written by engineers who know the system, it should be written by an end user without any assumptions - that's my opinion anyway.

regards

Robert

P.S. Klaus, Josef informed me that you were making equiries regarding the flash output on the D800 but Nikon are not willing to devulge any of their secret squirrel business - very sad. Thanks for trying.

[lumelix]

Re: Banding

by lumelix » Sun Feb 23, 2014 8:54 am

Hi Robert
Nice that it work's as suggested

As I wrote, there is allways vignetting on a f1.4 lens, even at f8 or f11. You couldn't see it in a single Image, but with a pano and blue sky, it's really visible.

Try to use the Standard Setting -2 for blending, should work also.

You can change a color anchor points back to one of the other settings, then it's not longer a anchor point.
Be carefully with more than one anchor point in the same pano! With multiple anchor points, there are other tedious blending problems arising in APG, as I wrote above and in many other posts.

Do you need informations about the Flash Controlling System of a D800?

[Boley]

Re: Banding

by Boley » Sun Feb 23, 2014 9:07 am

I purchased a Panoneed last year and it works perfectly except in the 2nd snc mode which is basically a signal from the camera flash sync to tell the head that the shot has been taken and to move it to the next location. We have isolated the problem to this area.

Klaus has the same head and it works perfectly with his D800 and even the voltage outputs are similar. Josef at Type&Colour was looking to get the specs from Nikon but they are not willing to devulge this information. Being on the other side of the world makes solving the problem a lot more difficult but I am sure we will get to the bottom of it sooner rather than later.

Robert

[lumelix]

Re: Banding

by lumelix » Sun Feb 23, 2014 9:14 am

Have you checked the cable and connectors?
Maybe a shorter cable can help also.
The sync protocol of D800 and D800e are the same.

[Boley]

Re: Banding

by Boley » Sun Feb 23, 2014 9:30 am

Replaced both cables and did discover that we had a faulty trigger cable which made the problem harder to find. The controller has also been replaced. I am hopeful that we are getting close.

[lumelix]

Re: Banding

by lumelix » Sun Feb 23, 2014 9:40 am

Which flash synchronism settings are you using?

[Boley]

Re: Banding

by Boley » Sun Feb 23, 2014 9:56 am

Tried both front and rear curtain

[klausesser]

Re: Banding

by klausesser » Sun Feb 23, 2014 12:40 pm

Thanks for doing the tests Klaus.

As a matter of fact i copied it from Ken Rockwell´s site! I´m a bit busy these days

best, Klaus

[klausesser]

Re: Banding

by klausesser » Sun Feb 23, 2014 12:52 pm

Tried both front and rear curtain

How did you set the internal flash? I never use it, but i set it to "TTL". But we think it´s either in the cables, jacks or something like that.
Joef told me that in a mail you mentioned to use a GPS added to the chain - Josef and i think there might be an issue regarding the common grounding . .
you might get different potentials which causes confusion . . .

Basically both cameras D800 and D800E are absolutely identical - aside from the AA-fiter. For me it works perfectly.

Can you precisely describe again your connections?

best, Klaus

PS: Aaron Priest uses the same head and controller - he has TWO PromoteControl devives chained on the head/D800 without issues.
http://www.aaronpriestphoto.com/2013/07 ... -the-lawn/

Last edited by klausesser on Sun Feb 23, 2014 1:45 pm, edited 2 times in total.

[klausesser]

Re: Banding

by klausesser » Sun Feb 23, 2014 1:08 pm

As I wrote, there is allways vignetting on a f1.4 lens, even at f8 or f11. You couldn't see it in a single Image, but with a pano and blue sky, it's really visible.

Right, Martin. Each and every lens in the world vignettes - more or less: that´s optics. But the vignetting of this 1,4/85mm Nikkor AIs is SO minimal above f4 that it´s practically invisible.
In my eyes it not at all can cause the banding which Robert showed us.

I´m so very much convinced of this because i have the same lens - and i dont´t have the issue at all shooting @f8 or 11. Never.

I had an issue with banding two years ago - bust that was the 1,8/85mm AI Nikkor @5,6. The 1,4/85mm lens has MUCH less vignetting. And even with the 1,8/85mm AI i usually have no vignetting @f8 or f11- as long as i use PTGui . . . or exessively use the anchors in APG.

So: what does this tell us? The - almost not existing visually - vignetting of the 1,4/85mm AIs Nikkor alone cannot cause the banding. That´s what i experienced.
It´s such a great lens - if THIS lens causes banding, 99% of all other lenses must be causing banding much heavier . .

Actually i use the lens twice a week and i NEVER run a vignetting-correction prior to stitching. I use PhaseOne´s CaptureOne for processing my RAWs . . . and i always have vignetting-correction deactivated.

best to you, Klaus

PS - @Robert: do you use a filter on the lens?

[Boley]

Re: Banding

by Boley » Sun Feb 23, 2014 2:23 pm

Hi Klaus, I have tried it with or without GPS and the problem still exists.

In regards to the lens, there is a UV filter on it, I wouldn't think that would cause it but I will try it without next time.

regards

Robert

[lumelix]

Re: Banding

by lumelix » Sun Feb 23, 2014 5:39 pm

Hi Robert
Just an idea, perhaps different Firmware than Klaus in the camera or the Controller ?
Did you use third-party accus in the D800e ? Or use it with the MB-D12 ?

[klausesser]

Re: Banding

by klausesser » Sun Feb 23, 2014 6:33 pm

Hi Klaus, I have tried it with or without GPS and the problem still exists.
In regards to the lens, there is a UV filter on it, I wouldn't think that would cause it but I will try it without next time.

Hi Robert!

Let the filter away - guess you fasten the sunshade on the filter´s ring, do you? Depending on how deep the filter´s rim is and depending on the sunshade this might cause issues. Rather unlikely with teles - but not impossible with small apertures.

I talked with Josef today - he forwarded your mail to me (hope that´s ok for you!?)

You wrote:
". . after I set the camera and head up ready to start the sequence I had one last look through the viewfinder and went to depress the trigger half way to check exposure when the head started to work on its own"

After starting the sequence, do not half-press the shutter!
Josef on my advise implemented the feature that needs a second confirmation for finally starting the sequence.
The reason is: when you finished all setrtings and push "start" - i realized it´s wise not to start immediately but have a last chance to look through the viewfinder to see whether there´s something in the way or so. It´s NOT thought to do measuring again!
Pushing the camera-release button half down then interrupts this process. I see no reason anyway to take a new measure - you measured already some seconds ago before you hit "start" the first time - or didn´t you?

I mean this "second start" option being the last chance to have a look and control the scene is very helpful - otherwise the head would start moving immediately.
This way it was before i asked Josef to implement this "second start option".

Another point is: this way you can connect a radio-control device for starting the sequence remotely after you did all settings and have the chance to hide somewhere in the case you can not move around behind the camera for being not in the pano. I had that sometimes in very narrow places.

So you hit "start" in the settings-menue, hide away somewhere and then push the radio-remote trigger.

Again: after having hit "start" in the settings-menue please do NOT take any measuring by pushing the camera´s release button half for measuring!

best, Klaus

[klausesser]

Re: Banding

by klausesser » Sun Feb 23, 2014 7:40 pm

In regards to the lens, there is a UV filter on it, I wouldn't think that would cause it but I will try it without next time.

Hi Robert!

Talked to Josef some minutes ago - he told me you needed a very long rail for your lens.
Can you tell me about the distance of your NPP on the rail´s marks with the panos where the banding occurs?

The 1,4/85mm´s NPP in my case is @ 70,5 on the standard rail. What is yours?
Do you take the long rail of about 20cm for THIS lens?

best, Klaus