Thanks to visit codestin.com
Credit goes to www.scribd.com

Scribd
Upload
54 views15 pages

Understanding Compositing Techniques

Compositing is the process of combining two or more images into a single scene. Key aspects of compositing include: 1) Using an alpha channel to store matte information and capture fractional occlusion between images. 2) Making assumptions about subpixel coverage and geometry when images are combined that approximate coverage based on alpha values. 3) There are 12 possible compositing operators that determine which images contribute color in different subpixel regions.

Uploaded by

prizmaweb
Copyright
© Attribution Non-Commercial (BY-NC)
We take content rights seriously. If you suspect this is your content, claim it here.
Available Formats
Download as PPT, PDF, TXT or read online on Scribd
54 views15 pages

Understanding Compositing Techniques

Compositing is the process of combining two or more images into a single scene. Key aspects of compositing include: 1) Using an alpha channel to store matte information and capture fractional occlusion between images. 2) Making assumptions about subpixel coverage and geometry when images are combined that approximate coverage based on alpha values. 3) There are 12 possible compositing operators that determine which images contribute color in different subpixel regions.

Uploaded by

prizmaweb
Copyright
© Attribution Non-Commercial (BY-NC)
We take content rights seriously. If you suspect this is your content, claim it here.
Available Formats
Download as PPT, PDF, TXT or read online on Scribd
You are on page 1/ 15

What is compositing ?

• Compositing is a technique of combining two or


more images into a single scene.
• Note that the individual elements are disjoint in
depth
• Compositing is associative but not commutative.
• We use a mask (matte) for each image that tells us
what portions make it to the final scene.
Pre-multiplied Alpha’s
• Normally pixels are represented as (R,G,B, α ).

• During compositing operations, we frequently find


that colors need to be multiplied by the alpha
value
• We could pre-multiply the R,G,B values with
alpha, pixel represented by (r,g,b, α) where r= α
*R, g= α*G, b= α*B
Use Alpha Channel to store
Matte Info
• We could use a 1 bit mask…but what if we
want to combine antialiased images?
• Anitialiased images have soft edges which
we want preserved in the composite.
• We use the 8 bit Alpha channel to store
matte information.
• This allows us to capture fractional
occlusion.
Assumptions about subpixel
coverage
• When two images are combined, we have
no information about coverage within a
pixel.
• Unknown subpixel geometry information is
approximated using the alpha value.
• If A is an opaque object with alpha value
αA , then A divides the pixel in two areas in
the ratio αA:1- αA
Example: Subpixel view with αA=.5

● ● ● ●

● ● ● ●
       

       
Likewise, an opaque object B with alpha value αB,
divides the pixel in two areas in the ratio αB: 1- αB

     
+
     
+
     
+
     
+

Example: Subpixel view with αB=.25


 
If B and A are both combined, then we assume that B
will divide each of the subpixel areas inside and
outside of A in the same ratio αB:1- αB

●+ ● ● ●
 

●+ ● ● ●
 

+      
 
 

+      
 
 

Result of assumptions

Region Area
covered by:

A and B αA*αB

A but not B αA(1- αB)

B but not A αB(1- αA)

 
  neither (1-αA)(1- αB)
   
 
 
 
 

 
●+ ● ● ●
 
 
αA(1- αB)
 
 
 
αB*αA
  ●+ ● ● ●
   
 

+      
 

αB(1-αA)
+  
 
    (1-αA)(1- αB)
This assumption is valid only for uncorrelated mattes!
Note that B does NOT divide A in equal ratio, because A
and B are correlated.
Here the background is completely covered, though we
expected .25 of pixel to show the background.

Background Middle (B) α=.5 Foreground (A) α=.5


Table of Subpixel Areas
 

Region Name Possible choices


covered by: that can contribute
to composite
A and B AB A,B,0

A but not B A A,0


 
 
 
 
B but not A B B,0
 
 
  neither 0 0
 
 

 
Compositing Operators

A binary compositing operation can be


defined as a quadruple indicating the input
picture which contributes to the composite
in each of the four subpixel areas.
• There are 12 possible quadruple
combinations. Thus 12 operations.
Composite Arithmetic: Alpha
for pixel
• Each input picture survives within its own matte
• For A, FA is the fraction of the matte that survives
in the composite pixel.
• For B, FB is the fraction of the matte that survives
in the composite pixel.
• Let alpha value of composite be αo
• Note that αo also represents the total fraction of
the pixel covered.
• Thus αo= FA*αA + FB*αB
Composite Arithmetic:
Average Color for pixel
• Average color for composite pixel=Color of
picture A, times its fraction FA
• +Color of picture B times its fraction FB
• Co= (αAFACA+ αBFBCB)/( FA*αA + FB*αB )
• co= αoCo= αAFACA+ αBFBCB= αAcA+ αBcB

You might also like