More Realistic Field of View for Panoramas & Default EXE

Please cast your votes in our two ongoing feedback polls here and here!

    Thought I'd take a moment and explain something if I may:


    Every camera - whether virtual or real - needs to establish a focal length/field of view. This is no different with a 360° image. Otherwise, you'd have no image, lol!


    I use the term FIELD OF VIEW in these cases because that's what 3D software such as V-Ray or Enscape typically use. Another term could be FOCAL LENGTH. While the two are distinct elements in the world of film and photography, they are connected and basically arrive at the same result:


    In conventional photography, whereby you are creating a full 360° pano image and stiching the shots together, a longer focal length/lower Field of View will bring objects CLOSER to the camera but require more stitching -because you've generated more still photos. With a 360° lens - or a render engine - what's happening with a longer focal length/lower field of view is that the same amount of visual data is present in the image, but the final output looks much more distorted (until it's unpacked in a viewer).


    When creating a pano, the Enscape camera is set to a to a very short focal length/higher FoV. In V-Ray and other engines, you can control this for the final output.



    Still waiting on this fix, by the way. ^^

    The player Enscape use to display panos gives visually ‘odd’ results, I use 3DVista to make my panos tours (eg https://www.tours.blockcpm.studio/percyplace/tour/, https://www.tours.blockcpm.studio/lancaster-pk/tour/, https://www.tours.blockcpm.studio/chubby-cherub/tour/) and the same panos look very different in the two.

    I only EVER use the Enscape viewer to quickly share a pano with my colleagues and only if it’s needed asap and there isn’t time to knock up a quick tour.


    If as you say landrvr1 its potentially down to how the pano is captured then I can’t understand how I get different results with exactly the same panos.

    @landvr1 - I would have thought that 360x180 degrees IS the FOV/Focal Length of the pano camera. I wouldn't imagine that the render engine is stitching together multiple shots like we would have to in conventional photography, but would love to know if I'm off base! This would put the blame on Enscape's browser viewer (very weird depending on aspect ratio), but not the flat images themselves.


    The office space comparison on Yulio from the OP is amazing, but it's worth noting the station points are different with Enscape being set farther back in the space than Vray. This makes it hard to compare apples to apples. I've tried matching views between Enscape and Twinmotion (also not adjustable) and get nearly identical spatial results with the same viewer. If you render out the exact same pano from Vray with different FOV settings, how do the images compare?

    Quote from landrvr1

    I use the term FIELD OF VIEW in these cases because that's what 3D software such as V-Ray or Enscape typically use. Another term could be FOCAL LENGTH. While the two are distinct elements in the world of film and photography, they are connected and basically arrive at the same result

    IMHO "focal length" shouldn't even be used w/respect to rendering. There is no lens, no sensor, no film. The only variables you really have are FoV and station point. "Pulling" objects closer by changing FoV does not change their apparent relation to each other, unless you also move the camera. FoV is very easily measurable by placing a couple reference objects in the scene, and is obvious in a program like 3dsMax that shows you the camera object and view cone. Since Sketchup, Rhino and Revit all handle cameras in obtuse (no pun intended) ways, a simple geometric principle is made unnecessarily convoluted.

    Quote from Backbeat

    ....This would put the blame on Enscape's browser viewer (very weird depending on aspect ratio), but not the flat images themselves.


    The office space comparison on Yulio from the OP is amazing, but it's worth noting the station points are different with Enscape being set farther back in the space than Vray.

    Agreed on this. It really is all about the playback platform. The difference in the example images is due to the station point - if it were due to FoV the difference between 80+ and 55 should be more noticeable. If the OP still has that example file I would ask them to regenerate the 360s in each render engine from the same camera point.


    The only problem here is by default the Enscape viewer shows "too much" of the image - ie too wide an FoV. Most other 360 platforms let you zoom in and out (effectively, reducing or widening FoV.) Escape's web viewer does not. And yes, aspect ratio is another variable. Some 360 platform "lock" the vertical FoV, others the horizontal. Compounding the problem is whether your audience is viewing on desktop (horizontal) or mobile (likely vertical.) For doing any comparisons I would suggest making your browser window square-ish to remove that variable.


    This is easily solvable by using any other platform to view the images, but if Enscape is going to promote their 360 tour capabilities they should address this. It seems like a trivial fix to reduce the default FoV and allow zooming in the viewer.


    W/regard to panos in VR - as you noted, that shouldn't be an issue. 360° is 360°. The FoV is determined by your headset. The topic of "stereo" panos is another discussion and this has gone on too long already.....

    Respectfully disagree. The camera for my examples was in the exact same location (created a point in Max where the camera was located and imported that via FBX into SketchUp, which I then used to set the camera location), and this is not purely a viewer issue by any means.


    The main problem still originates with the totally unrealistic FOV default in the Encape camera when creating a pano. The examples I originally posted speak for themselves, as they clearly demonstrate - without any adjustment on the part of the Yulio viewer - the differences in the original images.


    In fact, Yulio - like many automated viewer apps - gives you zero control over the software FOV unless you manually use your scroll button to 'zoom' the viewer image.

    Also, I would encourage others to try the same experiment with V-Ray or any other engine that allows control over the camera FOV. If memory serves (and it's been awhile) I actually recreated the Encape distortion quite easily in the V-Ray created image by simply increasing the FOV and dumping the image into Yulio.

    Quote from landrvr1

    The camera for my examples was in the exact same location.

    It isn't. Turn the camera 90 degrees in the pano. The VRay one is centered on the corridor. The Enscape one is backed into the curtain. In fact, since a lower FoV would bring objects closer to the camera - the wood wall should appear larger in the VRay version. It is virtually identical in size in the two images. The rug here is causing the Enscape one to feel 'deeper' but that is misleading, since the rug geometry extends further in the VRay scene.


    I've just knocked this up: https://www.tours.blockcpm.studio/00_sandbox/tour/

    • I rendered a single pano from a model.
    • I uploaded it to the Enscape cloud server and I saved it locally.
    • I used the local pano jpg in a 3DVista tour with it set to be 50% wide, 100% high.
    • I used the link for the cloud pano in an iFrame set to the same 50x100%.


    Rescale your browser window to get the aspect(s) you want.

    You can zoom in/out with the 3DVista pano.


    The most significant difference is the way they each handle scaling, the 3Dvista retains the width of the view whilst the iFramed Enscape cloud pano retains the height. This results in the 3DV pano looking better in a landscape aspect but the cloud pano looking better in a portrait aspect. Try it out by wildly rescaling your browser.

    I honestly think this is the cause of all our competing opinions; I view it in landscape and declare that 3DV is better and someone else views it in portrait and declares the Cloud version is better.

    Quote from rifkin

    It isn't. Turn the camera 90 degrees in the pano.

    Okay, you're correct! I must have uploaded the wrong pano. The camera is off by about 36" or so....


    However, my mistake actually proves my point even better.


    Here's a side by side of the initial view. Even though the camera in the Enscape shot is further back towards the curtains, the distance from the far curtain wall actually increases. The only way that happens is if the native FOV in the Enscape image is way off. There's no other explanation.



    I'm not disagreeing that the viewer can play a part in correcting this issue. However, adjustments on the viewer side only zooms the native image and it's quite apparent when looking around that the movement takes on the feeling of panning around the image. The more you have to zoom in to correct an FOV, the worse that effect.


    BTW, Focal Length has been an adjustable setting in V-Ray physical cameras since their introduction. The entire idea is to virtually recreate every aspect of a real camera. It's an absolutely valid use of the term in the 3D world.

    Quote from landrvr1

    However, my mistake actually proves my point even better.


    ...the distance from the far curtain wall actually increases. The only way that happens is if the native FOV in the Enscape image is way off. There's no other explanation.

    It does not prove your point better. The camera is further away from the curtain wall, so yes, the apparent "distance from the far curtain wall" increases. I will be happy to be proven wrong and further my own understanding, but unless/until we have a VRay vs Enscape render from the same camera point there is no relevant comparison here. I hope to set up an example scene but right now I'm too busy arguing on the internet ; P

    Quote from Paul Russam

    I honestly think this is the cause of all our competing opinions; I view it in landscape and declare that 3DV is better and someone else views it in portrait and declares the Cloud version is better.

    Thanks for doing this comparison. Can I ask, is the "+" in the center just for reference to align the two images? FYI the scroll wheel doesn't work if the cursor is positioned over the "+" so the zooming seemed a bit glitchy until I realized that's what was going on.

    The + icon is only there to give us a center if you want to line up on, say, a picture frame corner. It was whatever was available in the built-in 3DV library.

    I never tested the zooming whilst hovering over the icon but I get why it doesn't work .... I'm not going to bother changing it as I'll probably be trashing the lot when I next need to test something (I use Sandbox for my trashable tests)

    landrvr1


    I think their may be a misunderstanding of what FOV does.


    Changing FOV is *exactly* the same thing as cropping an image.


    If I render a 90 degree FOV, and crop it, I end up with *exactly* the same image as if I rendered with a smaller FOV that matches the cropped versions extents.


    A panoramic image works exactly the same. The viewer is just "cropping" (while also changing projection)


    The only thing that would fix your issue is the ability to choose a default FOV of the viewer (or use a different viewer)


    Nothing needs to happen to the panoramic image which is saved when rendering.

    rifkin and everyone, I yield and concede. My apologies. I've really misunderstood what was happening from the beginning, lol. I've no excuse. Carry on and I will now hang my head in shame,