CGI vs. Real Pics Test??

Many people cry "CGI" when someone posts pics on this forum. I see a lot of these cries in the UFO forums (of course). Since many believe that CGI is just as good as real pics and you can't tell the difference, it would be awesome to post a comparison test. I am not good yet of posting pics up on these forums.

If someone is bored enough and knows a lot about CGI, would someone want to take up the challenge to post 10-20 pics of CGI and real pictures intermingled together. Give the community a few days to view the pics and list if they are CGI or Real. Release the answers after X amount of days. This will prove if CGI can 100% pass as real pics and you can't be fooled by them. Kind of like the Remote Viewing Test Thread .

Hopefully, someone with posting pics "know how" will find organizing this test fun. In the mean time, I'll try to learn to post pics incase no one volunteers. Any bored takers out there??

Another reason I would like someone else to make up the test is I want to try it out myself. I think you can tell the difference between the two. I want to see if I'm wrong.

[edit on 2-7-2007 by Quazi176]

Agreed, I'd really love to see a test like this run by ATS. I'd rather not start it because I want to be able to test myself as well. How about somebody who knows how to do CGI?

Other sites have versions of this test already.

Heres one, there are many more.

Take the test, see how you do

Thanks Tiloke, someone else sent me another link here>>>CGI TEST.

I just thought someone could have fun doing a test on here themselves, but this is much easier. Thanks

Not to mention, a test on here, if everyone listed their answers, everyone can see who actually can tell the difference.

[edit on 2-7-2007 by Quazi176]

I got 8-10 right. Very good pics, but to me, there is just "something not right" to most CGI pics. Kind of like movies that show models as actual props in the scenes, they just have a certain "fake" look to them. It's very suttle, but there, none the less.

By missing two, proves I can't be right all of the time, but a pretty good %.

2nd Test 82%, 41 out of 50. I'll try one more test.

Ok can't find any other good tests.

[edit on 2-7-2007 by Quazi176]

[edit on 2-7-2007 by Quazi176]

There are a few questions on the CGI test that i think are real or fake or w/e but i still got them wrong but ive seen them elsewhere and it says they are real so you cant always trust just what some dudes test says if they are fake or real when he might not know for shure.

I see what you mean. My whole point to this was that, to me there is just something "different" about CGI and you can tell the difference. Not all of the time, but about 80%. From my experiments.

Maybe it's just a very slight change in the colors or the details. Just like picking a real pic over a pic of a model. There is a difference.

Originally posted by ParanoidKid
There are a few questions on the CGI test that i think are real or fake or w/e but i still got them wrong but ive seen them elsewhere and it says they are real so you cant always trust just what some dudes test says if they are fake or real when he might not know for shure.

But wasn't this about people on this forum making their OWN CGI to use in a ''CGI or real" test?

I doubt people would not know then.
And I also doubt the people at autodesk would not know the truth about those pictures lol.

Originally posted by -0mega-

Originally posted by ParanoidKid
There are a few questions on the CGI test that i think are real or fake or w/e but i still got them wrong but ive seen them elsewhere and it says they are real so you cant always trust just what some dudes test says if they are fake or real when he might not know for shure.

But wasn't this about people on this forum making their OWN CGI to use in a ''CGI or real" test?

I doubt people would not know then.
And I also doubt the people at autodesk would not know the truth about those pictures lol.

Yes it was, but a couple of sites came up (above) that have a test on them and no one has come forward to do this. I was going to try to figure the "posting pics" up here, but I'm still trying to figure out the [quotes] properly. I won't attemp pics until I can get other things down pat first.

The test above seems to do the job for now.

100% right in the first round of 'Fake or Foto,' but got 1 wrong in the bonus round. Even after I know it's a photo, that lizard still looks fake.

I know a little bit about photos. And I've got one here:

This was an image taken for an anti-drink driving campaign here in victoria. It's the basic image taken from the crash range. Do you guys reckon it looks real? Apparently it took em quite a few tries to get the pic they wanted, which is why theres so much stuff flying off the car

I got 100% in the first round of that 'Fake or Foto' test and 1 wrong in the bonus round. Dammit, even after I know it's a photo that lizard still looks like CGI.

Anyways, here's an old pic taken on the crash range of a company that runs anti-drink driving campaigns in Australia. It aparently took them quite a few tries to get the car to flip in the exact right place, which is why theres so much crap flying off the car. Makes it look more awesome, IMHO I've been told it looks like CGI. I don't see how. The window reflections line up with the shadow and everything. Why does it look fake?

The real standard should be, ok, if X photo is CGI, can an ATS'r duplicate it?

Also, we can have control photos, such as actual pictures of cows, etc., and see if ATS'rs can CGI those.

Wow, that pic of the car above looks fake to me. I know I'm not right all of the time, but that doesnt look real one bit.

www.autodesk.com...

First time, got 10/10 on this one. And 4/4 on the Bonus.

Guess I'm just good at that this sort of thing...

They still don't seem to be able to get faces right.

More important than guessing a bunch of pictures is knowing what clues to look for that CGI can't quite duplicate.

Most higher end graphic programs do well with reflections and shadows, but don't do well with fine structures, like hair and fur.

What do people look for in real photos that are composited, such as a landscape where someone pastes in a UFO?

One thing is to examine it in reduced Gamma, making the dark colors light. You can also do a 'negative image' and look for flaws. Sometimes inclusion of something small and odd mark a real picture, but if the pixelation looks smeared then it shows amateur attempts to make composites.

Another important thing is to have a full-sized graphic. It's pretty easy to hide compositing if you resize the picture after making the alterations. Anything smaller than 1024x768 and low rez will fool most people.

Also people can do paste ups, where they photograph a landscape with something pasted in. Then the photo is for all intents, real.

Originally posted by Chupa101
www.autodesk.com...

First time, got 10/10 on this one. And 4/4 on the Bonus.

Guess I'm just good at that this sort of thing...

Well maybe you could explain the clues that enabled you to guess correctly. Maybe it was just luck?

IMO, those photos are too small for good analysis, and they specifically try to fool you with pics that are normally CGI, like metallic objects.

The one that got me was the lizard in the bonus round. The shadow in the lizards right forward armpit shouldn't be there, as its head is very well lit seemingly by a light source that should also be shining on the armpit. Tell us how you got it, Chupa101.

Well, I clicked on the pictures so they opened up in a new window and were bigger, then looked for the tell-tale signs, like on the metal objects how good the texture was, the CGI stuff can't seem to pull off the 'metal shade/shape' if you get what I mean.

And yeah the coloured Lizard was the one that almost got me on the Bonus, but I just guessed that it was too obvious to be CGI (Luck I guess )

That's about it, just careful observations of the shading and textures.

Edit: Its does show though how advanced Picture editing can be now, I guess in 10 years Photo evidence will be impossible to prove/disprove.

[edit on 14/7/07 by Chupa101]

When I look at a CGI pic, there's just something not quite right about it. Maybe it's too glossy or the colors are not quite sharp enough. Whatever it is, you can tell be the first initial glance to a picture. If you have to stare at it, the better it is. There is always the one that slips by, but I've been able to tell 80% of the time. The CGI pics just have this fake quality to them.