Photo Gallery Forums

Recently I commenced using panorama plug-in and noticed some issues

1. After updating my spherical panorama with new version i see that to some extent old one is being loaded “under” the new one. I refreshed browser several times including clearing browsing data and the result is still the same. Browser Chrome on Mac BookPro retina display.

On iOs clearing the browsing data helps by the way.

2. Also noticed terrible banding of displayed panorama despite it is created from high resolution 16bit tiff file in marzipano tools. Not really visible on iphone but on desktop yes.

Link to mentioned panorama
http://kolasinski.art.pl/Galerie/Krajobraz/360VRN/

Any ideas?

Thanks in advance

I checked your example, and I see a CUBIC (multi-tile) panorama created from Marzipano. I also see that tiles look blurry as one navigates around panorama, which I assume is what you mean (by banding)? This is essentially how Marzipano works, as it will load appropriate level tiles (depending on screen resolution vs size vs density), and it will evict (unload) tiles which become invisible as you rotate the panorama, to free memory for the device.

This means that as you rotate, you will see high-res tiles loading on top of low-res tiles, even though the tiles are cached in your browser. I agree this is especially visible in your example, but it should work the same for CUBE panoramas in our X3 demo.

This was discussed earlier in this topic, and I posted a request to the Marzipano developers.
https://github.com/google/marzipano/iss ... -460973802

I believe this issue is resolved in latest Marzipano viewer, as they have largely increased the level of cacheable tiles. Please hold on until next X3 release coming soon, where we will include latest Marzipano, and I predict this will solve the issue!

Hi Karl

Not really, by banding i mean this phenomena https://en.m.wikipedia.org/wiki/Colour_banding

And yes, i’ve noticed blurness caused by progresive loading of cube faces and agree that it should be only once, not every time when viewer turns in certain part of panorama. Would be much better experience.

And again what about seeing part of previous version of panorama under the new version? Can You see it on you side? Try to zoom in and out as it is visible at certain level, depending on used device. Now i noticed this issue also on my iphone after zooming. You should see part of flying stork in the left part of panorama.

MaciejK wrote:Not really, by banding i mean this phenomena https://en.m.wikipedia.org/wiki/Colour_banding

This I cannot confirm. I can't personally see anything specific like color banding in your link ... Besides, if it's related to image colors, then this is related to the images/tiles. I can only guess something in the realm between color profile vs browser vs screen. The marzipano renderer can't create this effect ... It just loads image tiles!

Here is screen shot from my mac:

MaciejK wrote:On iOs clearing the browsing data helps by the way.

Not logical. Means you are loading different tiles after clearing the cache? Perhaps your device is doing something strange with colors to save memory? Some interpolation? Just a wild guess. It's not created by X3 or Marzipano, which just load your tiles.

MaciejK wrote:And yes, i’ve noticed blurness caused by progresive loading of cube faces and agree that it should be only once, not every time when viewer turns in certain part of panorama. Would be much better experience.

This will be improved in next release as the "tiles caches" will increase so that tiles don't unload and reload that much.

MaciejK wrote:And again what about seeing part of previous version of panorama under the new version? Can You see it on you side? Try to zoom in and out as it is visible at certain level, depending on used device. Now i noticed this issue also on my iphone after zooming. You should see part of flying stork in the left part of panorama.

What do you mean "part of previous" panorama under "new version"? I am guessing you mean that you see some artifact of one panorama scene under another panorama scene? A bit hard to imagine what you mean or what that looks like.

Also, what "flying stork" are you referring to? In the link you gave me, there is ONE panorama and I can't find any storks.
http://kolasinski.art.pl/Galerie/Krajobraz/360VRN/

Indeed on Your screenshot all looks just perfect, this is good thing. Really, really strange.
Here is what  i see on my mac:

Just to be more clear: there is only one panorama, right. The thing is that i first cerated one and uploaded to the server via FTP (it contained stork in the scene and had violet cast). I was not satisfied wit it and entirely deleted it by FTP, then created corrected version in marzipano and uploaded like before. Banding was visible already in first published panorama, then after replacing with new version issue with overlapping tiles arise.

Note red arrows pointing on more 'violet' faces which belong to first version of my panorama and part of the stork also belonging to first version  Also banding is noticeable on dark water and sky.

Take a look also on my header picture which has transparent background and only logo should be visible, without black around, this seems to be relevant issue:

Really mysterious why it is happening despite of refreshing page several times and clearing browsing data.

One idea, might it be due to quite specific internet connection i'm using which is satellite on the ship? But why? Some special newtork settings?

MaciejK wrote:Just to be more clear: there is only one panorama, right. The thing is that i first cerated one and uploaded to the server via FTP (it contained stork in the scene and had violet cast). I was not satisfied wit it and entirely deleted it by FTP, then created corrected version in marzipano and uploaded like before. Banding was visible already in first published panorama, then after replacing with new version issue with overlapping tiles arise.

Well that kinda explains it. In your case, you are experiencing CACHED tiles, and probably a mix of cached files from NEW and OLD sources. Surely that would be crystal clear from the fact that you see a stork, and I don't, and it would also explain "banding", in which case you are seeing a mix of old and new tiles. Tiles may be cached in browser AND/OR the network.

If you already had problems with banding in first version, I really can't comment on that, but you need to rule out the mixed version which you now clearly have in your cache, which would explain the issue. You could rename the exported project folder itself, which would mean urls would change to the tiles, and force refresh.

MaciejK wrote:Note red arrows pointing on more 'violet' faces which belong to first version of my panorama and part of the stork also belonging to first version  Also banding is noticeable on dark water and sky.

I don't see the point of discussing "banding" until you have solved your caching issue. "Banding" is not something that is produced by X3, Marzipano, or browser, which simply loads the tiles. In your screenshot, it looks like a mishmash of tiles from different exports.

MaciejK wrote:Take a look also on my header picture which has transparent background and only logo should be visible, without black around, this seems to be relevant issue:

Really mysterious why it is happening despite of refreshing page several times and clearing browsing data.

Works fine here. There are only two logical explanations: 1) You are being served cached content from network, or 2) You have some serious malfunction with screen or browser. I would definitely gravitate to #1 here.

MaciejK wrote:One idea, might it be due to quite specific internet connection i'm using which is satellite on the ship? But why? Some special newtork settings?

This is definitely not unheard of. We also have network caching because we use Cloudflare, which acts as a CDN, and will therefore cache all static resources. As a developer, I would never update a static asset without changing the name of the newly uploaded asset. Especially if you are on a satellite connection, it could be easily be using some additional cache mechanisms.