pacl A1 A2 A3 re-creations

[spf33]

          

really nice renders, hpo.

which software?  what kind of light and shadow did you use?
what kind of environment setup did you use?

curious because as i attempt some serious photoreal renders using hpo's excellent a1 model, which he graciously lent me, i'm finding it very frustrating to achieve the level of photorealism found in isaac's a1 jpegs. 

i'm getting some good results using vastly different lighting and scene setups, everything from a simple "domelight" using shadow maps on a black background with the a1 model on a white plane to a full on raytraced\global illumination\ambient occlusion\photometric lighting with the a1 model sitting a modeled light box.

but my private efforts are lacking in some photoreal skill when i compare my a1 renders (and hpo's, and lev's, and kris', no offense intended) to isaac's a1 jpegs - if looking at them as 3d renders.

[HPO]

really nice renders, hpo.

which software?  what kind of light and shadow did you use?
what kind of environment setup did you use?

Thanks,...but it's a very basic Mental Ray Max scene, with ray tracing enabled but without Global illumination.
The environment is just mr Sun with mr Sky, I also tried to make a lightbox but a simple scene got the best results.
To get the results as in the Isaac pictures with the supposedly black background and the greenish look, I've adjusted the original render with the simple mr Sky blue background to the renders you see above by adjusting the Hue to a greenish color and was surprised by the result, I also adjusted the saturation and lightness and put some gaussian blur and noise in the render.
I also tried different colors for the basic ground plane material.
 

curious because as i attempt some serious photoreal renders using hpo's excellent a1 model, which he graciously lent me, i'm finding it very frustrating to achieve the level of photorealism found in isaac's a1 jpegs.

i'm getting some good results using vastly different lighting and scene setups, everything from a simple "domelight" using shadow maps on a black background with the a1 model on a white plane to a full on raytraced\global illumination\ambient occlusion\photometric lighting with the a1 model sitting a modeled light box.

but my private efforts are lacking in some photoreal skill when i compare my a1 renders (and hpo's, and lev's, and kris', no offense intended) to isaac's a1 jpegs - if looking at them as 3d renders.

Your welcome,....and non taken, it's like I hear myself talking, it still misses something that is there in the isaac's pics, mine are to "sterile" but I haven't tried the printing and scanning of the renders so I don't know if that brings something.

BTW, do you have some results to show 

[Nemo492]

Spf, your turn... 

BTW I see that HPO uses 2 images (one small, one large)
I also used do do the same until I found this simplier code
(Please edit my Post to see it)

Zoom = right click then "show image"

[HPO]

Yes a nice code, but I like to keep it easier for the reader, although it's a little bit more work for me.

Photobucket also has a new thumbnail function, but it's a little to small in my opinion.

[spf33]

...a simple scene got the best results.

agreed. but i can't yet figure out a way to acceptably light the scene to get results that match the original jpg.  each method either introduces elements not seen in the original or is missing an element that is seen in the original.

we've discussed the lack any any specular highlights on the original a1 jpg and the counter argument has always been spec highlights are not seen because the a1 was photographed in some sort of lightbox. 

but i can't yet find a workable 3d solution putting the a1 model inside a lightbox model of any reasonable design and lighting scenario.

since this is a work in progress and i'm not nearly satisfied with the results and i don't want to junk up hpo's thread too much, here's a quick screenshot of where the file is right now.



in the center is the raw render straight from max, so no post work in photoshop.

and it's a very basic a scene setup; hpo's a1 geometry with a mr arch&design plastic material sitting on a white plane (an 8'-10' wide surface matches the reflections on the original fairly close using the known measurements of the a1 btw) with a black environment background lit with a mr_sun and mr_sky daylight system.

i leave it to the viewer to notice the differences between the render and the original.  but i will point out the specular highlights.

i've tried a few different methods to reduce\remove them while still keeping other desirable aspects intact, but at this point i'm thinking to match the original i'm going to have to render out the elements (shadow, diffuse, specular, etc.) separately and composite them back together in ps for more control.  but even then i'm not sure what kind of results i'll get as rendering out to elements and comp'ing in ps is never really part of my own work flow.

...mine are to "sterile" but I haven't tried the printing and scanning of the renders so I don't know if that brings something.

i've always thought that it would bring a lot.  now that i have your fantastic model to render with i'll definitely be doing some printing and scanning tests.

hpo, what do you think about the possibility that the a1 is a painted rapid prototyped model?  can you paint the polymer models to resemble the original a1 jpgs?

[HPO]

we've discussed the lack any any specular highlights on the original a1 jpg and the counter argument has always been spec highlights are not seen because the a1 was photographed in some sort of lightbox. 

but i can't yet find a workable 3d solution putting the a1 model inside a lightbox model of any reasonable design and lighting scenario.

Already a nice result spf !
 
Yes there is always a spec highlight in the renders that isn't there in the original, but the problem I have with the lightbox theory, is that in picture 4 (the above view you made the render of) you can see reflections on either side of what looks like a white groundplane to a darker wall or something.
So the lightbox isn't seen allready on two sides.

since this is a work in progress and i'm not nearly satisfied with the results and i don't want to junk up hpo's thread too much,

Yes maybe this topic should be split in two and lifted over to a research board.

hpo, what do you think about the possibility that the a1 is a painted rapid prototyped model?  can you paint the polymer models to resemble the original a1 jpgs?

Yes I think I can, but if this is a RP model it would also have been a staged photo, because of the floating components and it would collapse under it's own weight.

[spf33]

thanks for taking care of the split, hpo. 
i'm going to re-title the thread from recreations to re-creations, though.

now i'm looking for the other re-creations of the a1 device that have been done. 

any links to kris' or lev's please post or pm me the link.

[HPO]

i'm going to re-title the thread from recreations to re-creations, though.

Yeah,...my English 

[spf33]

got a chance to render the a1 scene from 3ds max out to elements in order to composite them back together in photoshop;



and this is the orig. a1 compared to the most recent 3ds max render with ps tweaks (900k):



i printed this approx. 1700x1100 render out on a somewhat crappy lexmark inkjet printer\scanner combo using 8.5x11 glossy photo paper;



i then scanned this 8.5x11 printed photo back into the computer using the crappy lexmark scanner at 300dpi:



this is after photoshop cleanup work, obviously looked much better before printing & scanning:



i thought for sure that printing then scanning the render would help immensely in the quest for photorealism\matching the orig a1 .jpgs.  but at this point i think i can say that printing\scanning had the opposite effect.

though on the other hand, this was the first run through of the method so i think now that i've completed the process 1 time closer results can be achieved.

but i'm finding this process frustrating.  still don't think i'm any where close to the realism seen in the original.  the original, if rendered cg, is filled with subtleties and nuances in geometry and lighting which are incredibly difficult to achieve in 3d and may be just be out of my reach as an artist to re-produce perfectly. 

but like i said this was my first real attempt...

things (re)noticed while re-creating the a1
- the surface the a1 sits on is minumum 6'x8'.
- lack of specular highlights seen in original difficult to reproduce in 3ds max without much extra work.
- printing and scanning so far does not help with adding realism to existing raw or ps tweaked render.
- the a2 and a3 objects are actually "in" the original a1 scene jpgs.  i.e. they
   are seen reflected in the a1 object and do not appear to be composited in.
- very difficult to get a close match to the original without tweaking the raw     
   render in photoshop.
- no camera or light diffusers in the reflections on the original jpgs.

[HPO]

Great research spf, I find it interesting that you get an opposite result then you expected with the scanning, but maybe a better scanner will give a better result, I don't know because I never use scanners.

Your animated gif clearly shows the geometry errors my model still has 
BTW did you lower the centre "crystals" for rendering purposes?, because they don't fit any more I noticed.

I'm interested in your follow up.

[spf33]

"...but maybe a better scanner will give a better result, I don't know because I never use scanners."

this is the scanner model i used.  i don't use it or scanners in general much either these days.  but i would assume it ranks among the lower end based on price\quality.

i was surprised at how much the color changed after scanning.  scanned the inkjet photo using the printer software which has no color controls and very limited brightness\contrast control.  also has a "pattern removal" checkbox which i think is apparent i left unchecked. 

also then surprised how much extra ps work was involved to get the scanned inkjet photo looking anywhere near acceptable.

but since i hardly ever use scanners i can't speak about the differences between scanning an inkjet on photopaper vs. a real developed-film photograph.

as a hoax are the isaac jpgs photoshopped to appear as scans - the hairs, the smudges?  seems like the easier route would've been to just scan rendered printouts.

Your animated gif clearly shows the geometry errors my model still has 

there are a couple, but all in all it's an incredibly well done copy. 

BTW did you lower the centre "crystals" for rendering purposes?, because they don't fit any more I noticed.

crystals?  what?

i made only one slight modification, moved the left front arm and circular foot\pad.  i slide it towards the left frame and bottom frame of the photo - down and to the left a little - to see what reflected.  on the original there is an interesting reflection on the rear of that foot and i was attempting to figure out what it was.  still missed getting it lined up properly though.

[HPO]

as a hoax are the isaac jpgs photoshopped to appear as scans - the hairs, the smudges?  seems like the easier route would've been to just scan rendered printouts.

You got a point there, why make it look like a scan when you can do a scan, only thing I can think of is, that it (in the hoaxers point of view) didn't bring the result he was hoping for when doing a real scan.

there are a couple, but all in all it's an incredibly well done copy. 

Thanks!

crystals?  what?

i made only one slight modification, moved the left front arm and circular foot\pad.  i slide it towards the left frame and bottom frame of the photo - down and to the left a little - to see what reflected.  on the original there is an interesting reflection on the rear of that foot and i was attempting to figure out what it was.  still missed getting it lined up properly though.

Well I had to give them some name  , but in the picture below is what I meant, the gaps are to big in those areas, like if the total of them are moved downwards, if you understand what I mean.

[spf33]

Well I had to give them some name...centrecrystals.jpg

nope, didn't touch any geometry except for what i described.
but i did use a mr_arch&design material with "rounded corners" on:



off:

on:


i might have over done the effect a little, but i don't think the setting i used is causing the issue you're seeing.

[HPO]

but i did use a mr_arch&design material with "rounded corners" on:

i might have over done the effect a little, but i don't think the setting i used is causing the issue you're seeing.

No I don't think so either, it's not a biggy of course, but how do they look from aside?, are the horizontal ones centred on those holes they stick in?
Otherwise I don't know what I am seeing  , because it doesn't look that way in the SW model, and in the renders I made.

[spf33]

Otherwise I don't know what I am seeing

ok, i'm an idiot.
guess i should have double checked the actual file instead of assuming:



good eye, hpo.
don't know when or how i accidentally moved the geometry, but it's definitely not misaligned in your original file.
not looking forward to re-creating the psd, but at some point i'd like to.