Anaglyphs - red and cyan glasses required to view.

124

Comments

  • AtlantisAtlantis Posts: 133
    edited July 2012

    Another optical illusion......

    And now i am off from this anaglyph thingy... I start already seeing strange 3d deformations and anomalies and impossible object's all around me ..lol :)

    platform.jpg
    1195 x 572 - 564K
    Post edited by Atlantis on
  • HoroHoro Posts: 6,539
    edited July 2012

    @shaykall - The idea is this. If you have the individual red and cyan pictures, you can superimpose e.g. the cyan one over the red one as a layer in a graphics program (e.g. Photoshop). Make the new layer 50% transparent and move it left/right until they match at an object that has the depth you wish to have at the screen plane.


    In general, look at a vertical line in the anaglyph. If the line has no fringes left and right, it is on the screen plane. If there is a red fringe at left and a cyan one at right, the object protrudes. If the colour fringes are inverted, the object recedes.


    Thank you for watching those videos without falling asleep. I have covered the anaglyph topic a bit differently a while ago on my website. Raytracing > Anaglyph? Perhaps it helps.


    Can you make an anaglyph from a picture? Not really. You need to have a separate picture for the left and right eye. I often do that in photography. I use HDRShop rather than Photoshop to make them red and cyan monochrome, than add together. If you use a graphics application like Photoshop, you would add then as layers and combine them. Looking at the histogram afterwards, you will note that you have only half the dynamic range (0 to 127). You have to brighten the resulting picture up. Saving the added pictures from HDRShop as BMP (or LDRI TIFF) retains the full dynamic of 0 to 255.

    Post edited by Horo on
  • shaykallshaykall Posts: 104
    edited December 1969

    Horo said:
    @shaykall - The idea is this. If you have the individual red and cyan pictures, you can superimpose e.g. the cyan one over the red one as a layer in a graphics program (e.g. Photoshop). Make the new layer 50% transparent and move it left/right until they match at an object that has the depth you wish to have at the screen plane.


    In general, look at a vertical line in the anaglyph. If the line has no fringes left and right, it is on the screen plane. If there is a red fringe at left and a cyan one at right, the object protrudes. If the colour fringes are inverted, the object recedes.


    Thank you for watching those videos without falling asleep. I have covered the anaglyph topic a bit differently a while ago on my website. Raytracing > Anaglyph? Perhaps it helps.


    Can you make an anaglyph from a picture? Not really. You need to have a separate picture for the left and right eye. I often do that in photography. I use HDRShop rather than Photoshop to make them red and cyan monochrome, than add together. If you use a graphics application like Photoshop, you would add then as layers and combine them. Looking at the histogram afterwards, you will note that you have only half the dynamic range (0 to 127). You have to brighten the resulting picture up. Saving the added pictures from HDRShop as BMP (or LDRI TIFF) retains the full dynamic of 0 to 255.

    @Horo. I will still need to watch the videos another couple of times, I am also reading the pdf transcripts along with the video's, (some things take a while to sink in),
    I will head off to your site as well .

    Thanks again.

  • shaykallshaykall Posts: 104
    edited December 1969

    Had another go :) will work on changing the target in the next attempts..

    MountainScene.jpg
    1514 x 885 - 181K
  • David BrinnenDavid Brinnen Posts: 3,125
    edited December 1969

    Couple more...

    B7_1_0_65_p13_s2_v2_anaglpherised.jpg
    750 x 750 - 1M
    Exspace_eg5_promo_anaglyph_m.jpg
    750 x 375 - 305K
  • shaykallshaykall Posts: 104
    edited December 1969

    Couple more...

    Very nice David, especially the canyon, almost got a nose bleed looking into it :))

    Still think the Sci Fi image you did was the best..

  • shaykallshaykall Posts: 104
    edited July 2012

    Had another couple of goes at this, but I have noticed that there is something not quite right at the bottom of the balls and cues on the table, some sort of blurriness, anyone know what causes that ?
    The first image I used the triangle as the target, the second was the left hand cue on the table and the third was the bottom left pocket (the one nearest us)

    I would love to do this scene to photo realistic standard but that means using the DTE and that is way over my head just now.
    I am going to follow Horos tut on the DVD and have a go at a previous scene but in color..

    PoolPocketANA.jpg
    1514 x 885 - 99K
    PoolTriangleANA.jpg
    1514 x 885 - 115K
    PoolTableANA.jpg
    1514 x 885 - 115K
    Post edited by shaykall on
  • David BrinnenDavid Brinnen Posts: 3,125
    edited December 1969

    Those look very good. The blurriness you see is caused by jpg image compression. Turn the compression off - as far as you can and things will look much better.

  • shaykallshaykall Posts: 104
    edited December 1969

    Those look very good. The blurriness you see is caused by jpg image compression. Turn the compression off - as far as you can and things will look much better.

    Hi David,

    I assume you are talking about the compression when I save the image after rendering the scene. I couldn't find anywhere to turn off the compression in Bryce so set it to max quality. this is the result..

    PoolTableANA.jpg
    1514 x 885 - 155K
  • David BrinnenDavid Brinnen Posts: 3,125
    edited December 1969

    Don't save from Bryce as .jpg save as .bmp and convert the image in a paint package where you have more control over the quality. I use Paint Shop Pro 8 which I got for £5 for a full licence.

  • David BrinnenDavid Brinnen Posts: 3,125
    edited December 1969

    And so on...

    B7_1_0_109_p33_s10_v2_anaglypherized2.jpg
    750 x 750 - 977K
  • shaykallshaykall Posts: 104
    edited December 1969

    shaykall said:
    Those look very good. The blurriness you see is caused by jpg image compression. Turn the compression off - as far as you can and things will look much better.

    Hi David,

    I assume you are talking about the compression when I save the image after rendering the scene. I couldn't find anywhere to turn off the compression in Bryce so set it to max quality. this is the result..

    ok Will give it a go :) thanks

  • shaykallshaykall Posts: 104
    edited December 1969

    Don't save from Bryce as .jpg save as .bmp and convert the image in a paint package where you have more control over the quality. I use Paint Shop Pro 8 which I got for £5 for a full licence.

    :) I am waiting for the special here in Australia for the option to get Paint Shop Pro 8 really cheap.

    I redid the scene, saved in BMP opened the images in Irfanview and saved them as Jpeg's converted the image to 3d and it looks really good, thanks, will be part of my work flow now .

    Thanks again

  • David BrinnenDavid Brinnen Posts: 3,125
    edited December 1969

    Don't let anyone say this lens system theory hasn't been tested...

    B7_0_1_26_p3_s8_v2_anaglypherized.jpg
    750 x 422 - 332K
  • HoroHoro Posts: 6,539
    edited July 2012

    Fantastic examples, David.

    @shaykall - you labour on your renders and wait until you have them in the best quality rendered. They are worth to be saved in a lossless format like BMP or TIFF. Your billiard table anaglyphs look very good.

    Post edited by Horo on
  • shaykallshaykall Posts: 104
    edited December 1969

    Horo said:
    Fantastic examples, David.

    @shaykall - you labour on your renders and wait until you have them in the best quality rendered. They are worth to be saved in a lossless format like BMP or TIFF. Your billiard table anaglyphs look very good.

    Thank you Horo, means a lot getting a compliment from one of the masters, I am going to have a go at an image in color..

  • David BrinnenDavid Brinnen Posts: 3,125
    edited July 2012

    This may help a bit with colour selection. This spectrum I've filtered for the Red and Cyan Channel. These show what is getting through for each eye. Then there is a monochrome section to show what the intensity is in that part of the spectrum that goes through to each eye. And the arrows show on the spectrum where the intensities best match for each eye. So for Red/Cyan colour choice the best colours to have predominant in your scene are yellow and magenta. The worst are, clearly, red and cyan. So if you have yellow and magenta they can be fully saturated without risk to odd effects within the eye, but otherwise if you have red and cyan, add some grey to the mixture so that they do not "flash". I hope the spectrum will make more sense than my explanation.

    Spectrum.jpg
    750 x 188 - 55K
    Post edited by David Brinnen on
  • HoroHoro Posts: 6,539
    edited December 1969

    And it does depend on the purity of the filters. With my cardboard goggles, I get the red left and right dark brown. With my new clip-on, they are pitch black. With left/cyan colour coding, red gives luminance, blue and green (cyan) colour.

  • shaykallshaykall Posts: 104
    edited December 1969

    @ David, I assume you are saying that with Red and Cyan glasses we need to avoid to much of those colours in the images and being better off with more Yellow and Magenta ?

    2 more images one in colour one just regular. they are night scenes with a small light duplicated, (I know I need to work more in the light lab:))

    VictorynightcolourANA.jpg
    1514 x 885 - 190K
    VictorynightANA.jpg
    1514 x 885 - 184K
  • David BrinnenDavid Brinnen Posts: 3,125
    edited December 1969

    Yes, I think that is the case. The proof of the pudding is in the viewing though, these theories are all very well, but a true test is just to experiment. Your black and white image works very nicely I think, your colour image seems a bit out of focus to me.

  • shaykallshaykall Posts: 104
    edited December 1969

    Yes, I think that is the case. The proof of the pudding is in the viewing though, these theories are all very well, but a true test is just to experiment. Your black and white image works very nicely I think, your colour image seems a bit out of focus to me.

    I agree about experimenting.

    your colour image seems a bit out of focus to me

    I thought so to but don't know why or how to fix it, they are the same images, with the same settings, and the colour with the center image added and calculated the correct way..

  • David BrinnenDavid Brinnen Posts: 3,125
    edited December 1969

    Ah... well I think that is now and old theory. The way forwards it so split the channels and keep the colours (but make them less saturated). So modify your source scene so your colours are less saturated and don't include any pure red or cyan. Then render for left and right as you did for the grey but filter the colours rather than set them monochrome. Then recombine the channels as you did for the monochrome. This is the process of the ALS we have made, but all the filtering and recombining is handled in the lens system in the scene so it renders it as an anaglyph. But that is essentially the process. All we have done is added automation. It can be done otherwise, it just involves more work and time.

  • HoroHoro Posts: 6,539
    edited December 1969

    Something went wrong with the coloured one, shaykall. Look at the mountain at far right. In the monochrome example, it has just a red edge, in the coloured one, there are two. The ship is ok in both examples, water and mountains are not.

  • shaykallshaykall Posts: 104
    edited December 1969

    Horo said:
    Something went wrong with the coloured one, shaykall. Look at the mountain at far right. In the monochrome example, it has just a red edge, in the coloured one, there are two. The ship is ok in both examples, water and mountains are not.

    Something went wrong with the coloured one

    Can't argue there Horo, I followed the tutorial on the DVD for doing Coloured Anaglyph.

    I used the small light amidships as the target;
    I centered the camera Origin X=0
    Rotation Y = 0
    Then rendered the scene as the center image, using the formula to set the camera for the left eye then the right eye I rendered the scenes. I then blended the images using HDR Shop, I have no idea what caused the pic to be so bad, seeing as the monochrome image seems much better.
    Maybe I'll try using another part of the scene as the target and see if that makes any improvement.


    @David I am no great CG Artists but will try to follow your advice about splitting channels and see if that also improves the image..

  • David BrinnenDavid Brinnen Posts: 3,125
    edited December 1969

    I should probably break it down a bit further. In the old method, you gather an image for each eye. Then can make it monochrome and then make it red for left and cyan for right and combine the two?

    Instead then. Before you render each eye, slip a 2D square in front of the camera. Make the material transparent so it has no effect on the scene and then set it's transparent colour according to which view you are getting.

    This should give you a left and right that is rendered red and cyan respectively (because the rays have to pass through an additional layer of transparency you might need to increase the maximum ray depth by 1).

    Now, take your two images and, depending on the software you have, in PSP8 "screen" them or in HDRshop use calculate and add the two together.

    The effect of this approach should give better colour results.

  • HoroHoro Posts: 6,539
    edited December 1969

    @shaykall - you certainly made everything right. It looks rather that there are 3 images added, not only 2. Look at the small peaks behind the ship and compare the two renders. The good one has 2, the funny one 3.

  • shaykallshaykall Posts: 104
    edited December 1969

    Horo said:
    @shaykall - you certainly made everything right. It looks rather that there are 3 images added, not only 2. Look at the small peaks behind the ship and compare the two renders. The good one has 2, the funny one 3.

    Hi Horo, I think you hit on the problem but I am not sure how to fix it, I will try to explain what I think is the problem

    The light amidships is the one I used for the target.

    1/ I render the scene with the Orientation X = 0 and the Rotation Y = 0
    2/ Using the formula I then set the scene for the Left eye X = -1 Y = 0.5
    3/ I then set up for the Right eye X = 1 Y = -0.5

    Using HDR Shop I can set up the monochrome anaglyph and it seems to be ok
    but
    When I add the center image to do the colour ANA I am getting a third peak,
    My theory is ( and please correct me if I am wrong)
    For the center render you draw a straight line from the camera to that mountain peak
    But in the Left eye render that line from the camera to that peak is coming in at a slightly different angle, the same with the Right eye render, the line to that peak is also coming in at a different angle so basically using the light amidship as the target it also becomes a pivot point and I am taking three snapshots of that mountain from slightly different angles.

    What do you think and how would I fix it, for colour ANA's should I use that peak as the target.

    @David, I will get back to you on your suggestions, going to setup now :)

  • shaykallshaykall Posts: 104
    edited July 2012

    I should probably break it down a bit further. In the old method, you gather an image for each eye. Then can make it monochrome and then make it red for left and cyan for right and combine the two?

    Instead then. Before you render each eye, slip a 2D square in front of the camera. Make the material transparent so it has no effect on the scene and then set it's transparent colour according to which view you are getting.

    This should give you a left and right that is rendered red and cyan respectively (because the rays have to pass through an additional layer of transparency you might need to increase the maximum ray depth by 1).

    Now, take your two images and, depending on the software you have, in PSP8 "screen" them or in HDRshop use calculate and add the two together.

    The effect of this approach should give better colour results.

    Sorry David, me again
    I set the scene for the left eye,
    created the 2d Square, put it in front of the camera
    went into the Material Lab clicked on the Transparent item in the Color menu and set it to red, then went down to the Optics menu and set all items to 0 except for Transparent which I set to 100,
    but when I render the image it has magnified the image, all I can see is the top of one of the sails, when I move the plane to one side the image is ok.
    Where did I make the hiccup?
    thanks

    Post edited by shaykall on
  • HoroHoro Posts: 6,539
    edited December 1969

    @shaykall - I see. Yes, the middle render as colour information. That was mentioned at around 9 minutes into the third anaglyph video on the Mentoring DVD. This is not a reliable method - sometimes it works, more often not. Obviously, in your scene it doesn't, my example in the video did - sort of. If you compare the monochrome and the one with the coloured one in the centre, you'll notice that the depth effect on the one with the coloured one added is not so good.


    The basis for those videos is removing colours by setting saturation to zero or using sepia tones. Monochrome anaglyphs work quite reliably. It appears that using filters when rendering retains a part of the colour. As long as there is no red in the picture, you get quite nice coloured anaglyphs, a bit low saturated, though. I discovered this when working on the anaglyph lens systems that permit the creation of anaglyphs right out of Bryce in one render. The camera looks through colour filters. If you wearing your red/cyan goggles and look around you, you can also see low saturated colours.

  • HoroHoro Posts: 6,539
    edited December 1969

    This is the Gremlin's Hall, an older render (see http://www.bryce5.com/details.php?image_id=4379 ) made up as anaglyph. Now I also undertand why it was so difficult to set up the lights.


    The toe-in angle is set for something very near to the eyes. Therefore, the eyes have a bit of a problem to focus. Also, the distance between the eyes is large, around double of an human. This makes the gremlins small (they are small guys, after all) and their cave huge.


    Best start at the lower left corner, where the landscape is nearest. Then move on to the waving guy, then continue right until you get the sitting chap fucused. At this time, foreground and middleground are easy to behold. Now move up your eyes and discover how vast this cave actually is.

    GremlinsHallV2ana.jpg
    800 x 600 - 260K
Sign In or Register to comment.