Sign in to follow this  
chewy0914

Life Like Graphics

Recommended Posts

Hello,

 

I posted this somewhere else.

Just found this site.

Thought I would post it on here as well.

-

-

-

Any character model- with maybe few changes.
Can be done.
So first thing first what makes a character in a game movie/other not look real???
take xbox(original) quality game??
What was the pixel resolution that a character could support?
Not the poly's the skin???
A few steps back Nintendo 64?
Mario 64- why didn't he look real????
Current characters????
Random starting point.
Take an Hd Picture of a person.
Zoom into the pixel level.
Look at the pixel distribution.
Look a the colors that are in the pixels.
Compare the pixel distribution to that of any character.
Look and see if there are any straight lines in natural pictures vs animated.
Look at the eyes. Are they white or tinted to the color of the iris?
took me 5-10 minutes to change eyes on a 512x512 .dds texture to make the eyes look real.
NFL 15 from the news is taking pictures of players to make models.
Do the look perfectly real? If no why?
Examples are:
http://www.123dapp.com/catch#d...
https://www.youtube.com/watch?...
What is the difference between the nearly life like cg in movies vs games as it apperars on your computer screen?
Both are pixels. Max res of your screen.
I'm
in the process of starting to put frames from a movie into autodesk and
make myself a skin for a model that will look like any high end cg from
a new movie.
If any game doesn't look right a point of
possible increase would be to take a photo capture of the game and
compare the pixels to what you want it to look like in a picture from
life or other animated material and look at the pixels. They Are The
Same.
Sp any game that doesn't look exact to life you
have your way to make it that way and if you can't you have something
that is already that way..
HD_Picture_person.jpg/raw/other

 

--

Anyone like portals??
They were able to take 2d pixels like any other game and make a hole in the game to another place.
You
would think if it is only pixels and you can see depth in a 3d
environment in a game you may be able to make anything on your screen
look like your looking through a door/window.
Doorway in a game vs a doorway in front of you.
3d movies whole movie setting in front of the screen?
Background/actors all.
I'll add more when i do.
--
Think cos-play to a whole new level.
You want to be batman....
There's the costume, there's autodesk, there's the character skin file.
I AM Batman.----------
--
If
any of you make cg movies for fun. try doing a screen grab of the
pixels or autodesking a character from a movie and see how close to a
Pixar or other movie you can get that will render in real time on an
xbox one.

-
-
-
I upped a res of a photo with image scaling it took some time....
Then I took the image back down to its original res.
The pixel quality was low.
Why that way????
I could zoom that higher res image out until it became a lower res image on my screen.
I haven't down it but I assume I could just zoom the whole image or parts down then do a screen capture keeping the high res quality in a lower res image.?
-
-
-
There is a way to make faster Gpu's now that use less energy and produce less heat.
Power usage/gflops mobile GPU vs desktop GPU
gtx 780m tdp 100
2528 gflops
GTX 780 250
3977 gflops
2x gtx 780m tdp 200
5056 gflops
3x gtx 780m tdp 300
7584 gflops
What would the tdp and how fast would a card be that has 4 mobile chips instead of 2
desktop one?

Edited by chewy0914

Share this post


Link to post
Share on other sites

Some sites didn't seem to understand how this could apply to all forms of visual media beyond games.


To clarify,
What causes something to look real?
form+color+more?

I took a dds files and took colors from images of real life objects
and applied them to the image in the dds file.
Replacing similarly toned pixels in the cg image with colors from the real life image.

It took about 20 minutes to do the ares I accomplished. After which the areas looked more
life like. I didn't change the model or the resolution of the image.
However I did increase the resolution later on with having the result of a cleaning character at higher and lower game quality settings. The change in resolution seemed to have no noticeable affect on the performance of my system.

What I decided from this there should be a means by which to make imagery character/scenes/other that are fully representative to life.

One example to which I noticed, as well as others may have is cg eyes. They don't seem to be real in cg characters. I was able to satisfactorily created life like eyes from pulling colors directly of the eyes of a high res picture of an actress.
When I extracted the colors at the pixel level I noticed that the whites in the eye were not what but a shade of the color of the iris.


So what I'm stating is that if the colors of life were sampled and those colors accurately applied
to a character/object or other the resulting image (3d/2d) would represent the coloration of which we call life. From examining the color pallet from a living portrait I believe the colors variation and base coloration do not exist as are applied to cg in life.


A low res picture of a person 512x512 pixels still looks like a person becasue of the colors
used in each pixel. If you up the res to 1024x1024 the person might look clearer but
the person is still the same amount of the picture.
5x5in at 512x512 and 5x5in at 1080x1080 person in both pictures is 3x2 in.
You most likely will be able in both to easily tell it is a person and it is the same person
in both pictures.


Pulling colors textures from the environment like autodesk which may or may not do it well
could be a way to get the look it seems games keep striving for.
(I took a picture of real steps and replaced the steps image and the steps that were
put in looked real.


This began me wondering all we see on the screen is a set number of pixels
1080p or that of your screen. A nice animated character in a movie such as The Teenage Mutant Ninja Turtles is really only the number of pixels on your screen as the max it can support. Regardless of the resolution that the animation studio created the model in. That could imply I could be able to remove frames a movie and create for myself a model of that character.
After which I found the software autodesk which may be able to perform that task.



-
With finding all of this it makes my question with those in the field who deal with such all the time why haven't they been able to discover that and apply it.
It would seem when a magazine does the changes of cg to a model they can do it so that real life is
added to with real looking cg.

With forth with the texture level that any system can handle if the colors were altered to life pallet would that not make the cg object look real and possibly better as a whole?

One example being I was working with an image of Laura Croft from the new game.
If the color pallet were altered to solely life found colors how much more real would the character instantly look? Forgoing that shaping of the characters figure would be needed as well.
Which may be easily accomplished by applying pixel distribution of a living person to the model.
Or combinations there of.



Other random thought. 3d currently can be produced 3d inside of 3d on a screen.
A door you can look through in a game looks as though you are looking through a door in real life being retaliative to the view of the character.

This being solely represented in pixels which are placed in a 2 dimensional row. Is it not that 3d viewing should be producable by means that allow viewing of 3d in the dame manner of the person looking at the screen?
rows that are 3d why is it that 3d coming out or into the screen is not being down to? It seems to be doable in another environment. It should be so that we could look
at our screen and see that we were looking down a tunnel. Or that the entirety of the game is outside of the screen. One example be a portal in a portal in the game Portals.

To clarify more if life accurate images are the goal, That has already been accomplished by
any picture displayed digitally.
And it could go that if that accuracy is the goal the only source needed with others being possible
could be a picture of life ,as well if need be multiple others, to extract the symetry of life from a source of pixels. Those provided the exact shape formation and more of the object wanting to be produced.

As for the way zooming in or out of an image seemed to affect the quality,
I submitted that for those who may have a better grasp or means of which I don't to test to see if that actually allows a higher quality to be produced in a smaller format than by other means.
As well as a possibly faster way to shrink an image than the current means of processing the image down to a smaller size.


Lastly noted: the portion showing the hardware was a side note instead of making a second
post comparing the performance of multiple mobile gpu's to a single desktop gpu.

The figures seem to show that possibly 2 mobile gpu's on 1 desktop card would use less energy produce less heat and provide more computational power than a single desktop gpu.
When then made me wonder that if 2 desktop gpu's can fit onto one desktop card.
What would it be if 4 mobile gou's were put onto one card?

Share this post


Link to post
Share on other sites

wat?

 

I think you should reduce the amount of words in your posts an start to make any sense with them. Really, I can only force myself to read about the top 1/3 of your posts before the confusion caused by your text gets too much for me.

 

 

... sorry, couldn't resist. Anyway: What exactly do you want to discuss here?

 

I guess the topic is about real life and somehow fake looking characters in contemporary and past games, and what makes them look real versus fake. Basically what you talking about goes into the direction of the "uncanny valley"... where something becomes so real looking that any little thing off is much more severe than with, for example, a cartoon character.

 

As far as I have seen it, there are multiple things playing into it:

 

- Animations that are off (to stiff, to limited, somehow wrong otherwise)

- Shaders that are not replicating the material correct (can be seen with metal extremly well, but skin is another material that is hard to get right. There is a reason why Toy story was one of the first big animation movies. Plastic must be one of the easiest materials to write a shader for)

- Lighting (fake looking lighting makes everything in the scene look fake)

- Missing highfrequency details because of missing bump map (this is much less severe than the top three if you ask me).

 

Now, the amount of polygons plays only a small role IMO... usually you will seldom notice if something is really lowpoly, as long as it has a good shader and a bump map attached to it. It can be a problem if an object needs to be simplified to fit the poly budget though.

 

Colors do play a role, but only for some materials (like skin)....

 

 

The gibberish about GPUs I really cannot decode... is there a question or a statement hidden in there? Is it some kind of a riddle?

 

I guess you want to discuss if a better GPU would be possible today:

What happens if you double the amount of shader cores (if that is your question) seems pretty straightforward to me. More power, bigger chip, more heat, lower production yields, therefore much higher price.

 

Even the scaling will not be 100% as there are a lot of other bottlenecks in a GPU (letting the CPU and Drawcall bottlenecks aside for the moment). As much as I love the big fat cards from Nvidia and AMD, there is a reason for the price and heat. So bigger is not always better.

Edited by Gian-Reto

Share this post


Link to post
Share on other sites


So what I'm stating is that if the colors of life were sampled and those colors accurately applied
to a character/object or other the resulting image (3d/2d) would represent the coloration of which we call life

To make it short: this is a common way of artists to choose the basic color palett, by picking colors from real-life reference images.

 

Nevertheless, this is only one part of (photorealistic)/life-like models. Do you know basic models, where they took some photo of a person and render it 1:1 on the face of a 3d human character ? Well, do they look life-like ?

Share this post


Link to post
Share on other sites

It's a lot easier to make a still image look life-like.

 

A lot harder when you add animation, not only get the motions right, but also get the (change in) lighting right.

 

Most 3D artists already use photos as a basis for their textures, but there is a lot more to it then that.

Share this post


Link to post
Share on other sites

Ill be honest, I don't even know what you are trying to ask. There is a lot in your message and most of it isn't very clear as to the "direction"

 

What are you trying to ask? Be specific if you don't mind! Lots of us want to help you find the answers to your question or get involved with the discussion. This is hard to do without fully knowing what your purpose/question is.

 

As for realistic rendering, most often today people are moving to physical rendering methods to create photo realism. This relies more on shaders than it does the actual artist.

Share this post


Link to post
Share on other sites

Create an account or sign in to comment

You need to be a member in order to leave a comment

Create an account

Sign up for a new account in our community. It's easy!

Register a new account

Sign in

Already have an account? Sign in here.

Sign In Now

Sign in to follow this