Tradeoff

Resolution Trade-offs
Lab Objectives In Lab 3.1 and 3.2 we explored the effects of changing quantization, which determines the number of bits per pixel, and changing the sampling interval, which changes the number of pixels in an image. Reducing either the number of pixels or the number of bits per pixel reduces the storage required for an image at a cost of lowered image quality.

In this laboratory we will explore how to stay within a "bit budget" for an image by adjusting both the quantization levels and the sampling to keep the best image quality. This can be important when deciding what images will be put on a web page or sent electronically

Verify that the image of the baboon on the right is the same as the image on the left and that the same number of bits is used for each image. || Remember to **zoom in** on the downsampled image so you can view it at the same size as the original image ||
 * 1. || Start the lab 3.5.1 Resolution Trade-Offs Grayscale. An old world monkey should appear. ||
 * 2. || Set the pixel size slider to 1 and the bits per pixel slider to 8.
 * 3. || Set the pixel size to 2.
 * || Q1- How many bits are now used in the output image?

131072 What is the ratio of bits in the output image to bits in the input image? 1:4 || What parts look different? fur, whiskers
 * || Q2- What parts of the output image look the same as the original image? nose, eyes

How would you characterize the parts of the image that show the most loss of quality? unclear || Adjust the bits per pixel until you have an image using the same number of bits that you recorded in Ql || Depth and fine lighting Which image is better - this one or the one from Step 3? Step 3 || Slowly reduce the number of bits per pixel until you just notice a further reduction in image quality || 3 bits per pixel How many bits are needed for the image? 49152 bits || In most cases the best quality will not be very good. Maximum bits in image pixel size bits per pixel 300,000 |-1 |-4 200,000 |-1 |-3  150,000 |-2 |-8  100,000 |-2 |-6  75,000 |-2 |-4  50,000 |-3 |-6 || Would our result be any different if we reduced the number of bits first and then downsampled? Yes || || 2 and 8 Is your choice different from the baboon image? No
 * 4. || Return the pixel size to 1.
 * || Q3-How many bits per pixel are used? 2 ||
 * || Q4-What loss of image quality do you observe?
 * 5. || Set the pixel size to 2 and set the bits per pixel to 8.
 * || Q5- How many bits per pixel are you using?
 * || Q6-For the following bit budgets, find the choices for each slider that will give you the best image quality.
 * || Q7- In this lab we are downsampling first and then we are reducing the number of bits per pixel in the downsampled image.
 * 6. || Now we will look at a different image. Load cityLights256x256 ||
 * 7. || Compare the output image with a pixel size of 2 and 8 bits per pixel to the output image with a pixel size of 1 and 2 bits per pixel
 * || Q8-Which image has the better image quality?

Why? Because it looks the best || In most cases the best quality will not be very good. Maximum bits in image pixel size bits per pixel 300,000 |-1 |-4 200,000 |-1 |-3  150,000 |-2 |-8  100,000 |-2 |-6  75,000 |-2 |-4  50,000 |-3 |-6 || Because you want more detail in different areas || What value of pixel size and bits per pixel did you use? 2:8 Would this image be suitable for a small part of a web page? Yes, the quality is not bad. Would this image be suitable for a full screen display? No, it would be distorted and stretched beyond recognition Use the same settings for the sliders that were used in Question 11 || It looks better when using the same settings How many bits are in this output image? 524288 || Becomes pixelated || It has more detail in the bushes and curves on the canoe What characteristics of the image content of the canoe image require more bits to be reasonably reproduced in the output image?
 * || Q9-For the following bit budgets, find the choices for each slider that will give you the best image quality.
 * || Q10- Why would you spend your bit budget differently on the cityLights image than on the baboon image?
 * 8. || Change the input image to the girlOnWater256x256.bmp image ||
 * 9. || Explore the effects of changing the pixel size and the number of bit per pixel.
 * (insert image here)** ||
 * || Q11: Find the best settings for an image with less than 250,000 bits.
 * (insert image here)** ||
 * 10. || Change the input image to the canoeBW512x512.bmp image.
 * || Q12- How does this output image compare to the previous output image in quality?
 * || Q13- What happens if you try to reduce this image to less than 25,000 bits?
 * || Q14-Why is the canoe image with less than 25,000 bits so much poorer in quality that the girlOnWater image?

What makes false contouring more pronounced? ||
 * 11. || Stop the lab. Close the lab ||