Static 35mm Adapter Solution

[Louis Feng]

Here is for my Sony PC101,

690K effective video pixels

that's 830x830 pixels, which is bigger than 720x480, but I think they use the extra pixels for stablization etc.

CCD chip width = chip height = 1/4.7 inch
= 0.212766 inch = 5.40426mm (1 inch = 25.4mm)
=5404.26 micron


size of each pixel = 5404.26/830 = 6.50596 micron

It may seem logical to concluded that 3 micron ground glass should appear to be pure white to the camera, but that's not the case. It's very close to that. I think there are more complex factors involved since the image to CCD translation is not 1:1. The actual image is much larger, and the lens shink it to that 1/4.7 inch area.

One of the problems is the little surface relief of the coarse surface can create shadows and reflect lights. In certain conditions these little shadow can become big enough to be visible. Also the reflections at some points of the surface can create small "sparks" that's brighter than its neighbors. Even if its smaller than 6 micron, it still contributes more light than its neigbor pixels. All of these little problems becomes less obvious when the ground glass is farther (a few inches) away from the camera.

[Nicholi Brossia]

Just to let you know in advance, I think I know what I'm saying... but I don't think I know how to say it. So please forgive my babbling.

The goal is to fit as much of the projected image into the video frame as possible. When you move the glass far enough away from the camera for the grain to disappear, don't you then have to zoom in to fill the video frame with the image? Since its an optical zoom, wouldn't zooming in look identical to moving closer, and thus the grains would appear to be the same size?
The grain size vs. pixel size calculation does make sense, but you have to take into effect how many grains are visible in the video frame (zooming in and out will effect the orientation between grain size and ccd pixel size). My understanding is that it isn't so much how many grains per area of the glass as it is the number of grains per area of the projected image relative to the ccd area. Both a 35mm and 25mm projected image would be identically composed on the video recording due to filling as much of the video frame as possible. However, using the same 3 micron ground glass, a 35mm image would appear to be less grainy than a 25mm image due to the 35mm image containing a higher concentration of grains.
Instead of calculating the 3 micron grains against the 6.5 micron pixels, it would be more accurate to calculate the number of "grains" in a 35mm x 25mm area (when viewed full frame) against the number of pixels in a 5.4mm x 5.4mm area ccd (which provides the actual full recorded frame).
That would be difficult. Technically we're not dealing with an exact grid of particles (grains) as much as tiny scratches and gouges in the glass. Honestly, I have no idea how to calculate scratches because they're so absolutely random. I considered calculating each grain as 3 micron x 3 micron (9 square micron) but don't think that would provide a realistic number due to the nature of the glass texture.

Hopefully that made some sort of sense.

Honestly, I'm still hoping for the best with Vendible's non-grain solution. Considering the upcoming market for higher resolution 720p camcorders, it would be a shame to have to totally reconstruct the adaptor to accomodate new camera's ability to see more detail and thus more grain. It sounds like, so far, you've been able to distinguish the grain on the on-board lcd/viewfinder which aren't very high resolution. If grain is visible there, it will most likely be even more pronounced on a 500 line resolution monitor.

[Louis Feng]

I probably should have mentioned that since I still don't have a housing for the setup, for most of my tests I put the GG right in front of the lens of the camera (literally touching each other). In this case the grains are visible. However, since my camcorder has a small lens, even when fully zoomed out, it still needs to be at least 2 inches apart from the GG to view the full 24x36mm image area. Whether the grains are still visible or not in this case I can not tell for sure untill I can have them fixed in a box.

On another front, I have figured out how to fix the hot spot problem. I know people used fresnel lens. But I understand the theory now and what it takes to have it completely removed. So I feel much more comfortable with more tranparent (less coarse) materials. That also means if the 3 micron GG doesn't make visible noise at 2 inches away, then we should have solved all the technical problems, and I can start prototyping it on computer and building it. If the other materials I'm getting work better than GG, then we are going to have something superb.

I know it only makes things more complex, but a condenser lens is also needed in addition to the fresnel lens to completely remove the hot spot. Here is some reading if anyone is interested: http://topcontechnotes.home.att.net/...tem/index.html.

If there is one high power fresnel lens exists (I don't know how much power yet), then we may not need additional condenser lens. If not, then condenser lens is needed to completely get rid of the hot spot. To get really high quality you can go with double lens etc. But there are math must worked out here to make sure we'll buy the right lens with right focal length etc.

I bought rotary tools that can cut and drill. Won't be here till a week later. I plan to build my first one with hard board and super glue.

[Louis Feng]

Ok, with a fresnel lens and a magnifying glass, I'm able to completely remove the hot spot. I'm still not sure what focal length the condenser lens should be.

[Noah Posnick]

I can't wait for this to be perfected since it is much simpler and more manigable in size. Since the focal length must be adjusted using distance of the lense I thought some sort of system with a pull focus like nob could work. Maybe we could adapt something like in this rig http://homebuiltstabilizers.com/stop%20motion%20camera%20dolly.doc

[Nicholi Brossia]

I was thinking the same thing, Noah. The ability to fine tune the ground glass distance would basically be like adjusting back focus. Honestly, the picture is too small to accurately eyeball-focus it on the ground glass. The best focus won't be available until the adaptor has been assembled and the image is viewed on a high resolution monitor. The screw method would provide an excellent external adjuster.

As far as the 2 inches away thing... that's great. That means the picture is covering a larger number of "grains" and thus less will be visible on the video picture. I was afraid that the image would be so small that, even with the ground glass right up against the camcorder lens, it wouldn't cover the entire video frame. I have a GL2 which sees a 4cm x 3cm frame when placed right up against the lens - so I'll have to get a close-up lens to get a proper shot of the ground glass image.

The website you listed is great. Its about time someone actually explains how a fresnel works rather than just saying it works. I had a feeling that the ground glass (actually plastic) on my SLR camera is ground on one side, fresnel on the other because there was such a noticible difference in image quality when I flipped it over (could have also been due to the slightly longer distance from the 35mm lens to the actual ground glass face?). Are you considering getting a real glass fresnel from Edmunds and grinding the flat side with 3 micron? If that works, that would be a great space saver and you wouldn't have to align nearly as much.

I definately like the idea of a double condenser lens and by-passing the need for a fresnel. I'd like to preserve as much detail as possible, again, considering the 720 and 1080 resolution (NTSC) cameras that are being released. I will read the information on that website again and try to use the formulas to come up with a proper combination of condener lens. I haven't done any physics related projects in years, so I may not accomplish anything... but I'll try.

Another thing I wanted to mention... on the SLR camera that I took apart, there is a thin piece of metal (almost a frame) between the ground glass/fresnel and the condenser lens. Basically the condenser lens stands just slightly off the ground glass. Just thought I'd mention that.

[Nicholi Brossia]

Yeah... its been way way too long since physics class.

[Louis Feng]

The fresnel lens is used only to reduce the size of the condenser lens. So it can be replaced by pure glass lens if size is not an issue. Since you have a condenser lens, could you measure the thickness of the lens, and post it here? I'm guessing without the fresnel lens, you need two pieces of 6mm thick plano-convex lens, each with focal length of 100mm.

[Filip Kovcin]

i just thought about something which is suitable for GG but not mechanically added on the surface, but organic. on the agus 35 thread there is an idea to use eggnog "film" to cover the GG, but maybe there is something more controlable on the market?
some time ago i saw on the discovery channel that some guys are producing kind of "jelly" glass which is controlable by (if i remember that properly) small amount of electricity. the key issue there was not the transparency, but the thermo isolation etc. (those glases will be used in homes instead of mormal glass)
BUT, there were examples of how that "glass" behaves - and in, say "state 0" it was fully transparent, and in "state 1" -semi transparent - "milked".

so, to make long story short - maybe there is another solution for that STATIC GG in organic world around us...

(since the atoms are probably smaller than pixels - joke:))

just a thought.

[Nicholi Brossia]

Okay, I just made a quick measurement of the condenser lens from my SLR camera. So far, I've made the most accurate measurement I can with a ruler (not very accurate), but I will get some calipers tomorrow to get a precise measurement.

The lens is 35mm wide x 24.5mm tall, with a 1mm "standoff" area between the flat side and the curve. The deepest part of the curve (center of the lens) is approximately 3mm.

Here's a quick drawing of the lens with measurements:
http://f1.pg.photos.yahoo.com/ph/nic...=condenser.jpg

I have some calipers, I just have to figure out where they are. I will find them tomorrow and post more accurate information.

[Nicholi Brossia]

Something else I've noticed with my SLR is that in the center of the ground glass, there are two concentric circles. Of course, these areas aide in focusing, each providing a more detailed/precise focusing area. The inner most area is by far more detailed than the outside, but makes diagonal lines look jagged, much like the "stairstep" appearence of interlaced video on a computer monitor. This is due to some type of coating/film used as a substitute for the grind. After viewing the area with a magnifying glass, I can see that the film is made up of numerous tiny "cells" almost resembling honeycomb... hence the jagged/pixelized appearance. I don't think that would look good on a television.

The middle area provides slightly less detail for a slightly softer picture, yet the diagonal lines are very smooth and natural. This provides an acceptable image.
The outer area is by far the worst quality both in detail and brightness... unacceptable in my opinion.

Okay, so what does this lead to? Exactly what was pointed out earlier today considering fresnel vs. double condenser lens. The innermost circle is the only one that contains the honeycomb looking film that I spoke of. The other two areas have been ground identically (same grit). However, the fresnel is only existant in the outermost layer (I can see the concentric circles with the magnifying glass... they're tiny) where the detail is much worse. The catch is that the middle area contains a very small bubble in the glass... in other words, the very peak of a second (technically the first in series) condenser lens. This tells me that the only acceptable layout is a double condensor lens. This will get rid of the hotspot (if you set it flat side down on a piece of paper, you can actually see it working... there is a shadowed circle in the center of the lens) AND sustain the most accurate detail. Honestly, there is a huge difference between the quality of the two.

If everything works out right (considering I know nothing of mother nature) we should be able to derive the focal length for two identical, thin condensor lenses to be aligned in sequence after the ground glass.

[Nicholi Brossia]

Here's a drawing I threw together to illustrate that last post.
http://f1.pg.photos.yahoo.com/ph/nic...dnm=SLR_gg.jpg

[Filip Kovcin]

hello there,

when i read about the solution WITHOUT GG, i was very sceptical, but anyhow ...
i just checked the option WITHOUT moving parts and WITHOUT GG.
here is the list of things used:

- camera sony PD 100 (52mm front filter diameter)
- macro lens +10 (single lens) from Hama (52mm diameter)
- cake CD box (for 50psc)
- zenit PHOTO camera lens 58mm/f2.0

i zoomed into the zenit's lens picture to avoid vignieting and that's all. no prism, no nothing exept macro lens with 10x power, and cake box with zenit 58mm lens on it. so it looks really simple. i did the tests with the camera, but cake box was not so stabile, so i kept that in my hands which results in vignieting, but this is due to my shaky hand. focusing is in this case really difficult. i changed focus here and there to see how it looks.

for me it looks really good:
- it HAS proper DOF,
- for some reason, colors are bit different (read: subjectively better)

also, because of low light conditions (just one table lamp on the begining and two ceiling lamps (2x 75W) + christmas tree lamps) i used 18dB image gain BUT with -3dB offset in main menu settings).

these gives me interesting video (grain) noise, which probably helps to see everything more "filmic".

maybe i'm too subjective, so please tell me - do you see what i see - or this is just my illusion?

maybe i'm fascinated with it, but to me it has something common with proper film feeling.


i can send my "filmeo" (film+video) tests today to you via taylor (with his kind permission) if you want to see it. should i send them?

[Zac Stein]

excuse my ignorance, but wouldn't just zooming into the back of another lens cause the 7.2 magnification we are trying to avoid?

Zac

[Nicholi Brossia]

I agree with you Zac, but just in case, I've posted an experiment on Agus's thread that hopefully others will participate in.