Artefact Four

For my fourth artefact, I rendered an image from the 3D model I produced for the Client Project. This had to show an improvement on the last artefact so I made a high quality single frame render of my redesign of the cocktail bar, ‘Minus One’. This image shows a well lit view across the bar including a realistic raised seating area.

I put the image into a questionnaire of the same structure alongside three photographs. Participants were asked to pick out the computer-generated image and to give any reasons for their choice.

The results showed that out of the people who participated, a modest fifty five percent of them correctly identified photo D as the computer-generated image. Most participants gave very limited comments when explaining their decision. However, a lot of people noted that it was a hard decision to make or that they were torn between two particular images. These findings show that there is an improvement in standard from the image used in the previous artefact. A much larger group of people were fooled by this questionnaire, which shows that a fairly decent standard of reality can be created on an inferior machine. Even though the percentage is only just a majority, the results are encouraging.

Client Project - Final video..

the quality needs to be inproved..

Artefact Three

When I put together my third questionnaire I created a 3D model based on the flat that I live in. I felt it was important to experiment with a visualisation that I had created myself on a machine with a much lower specification than industry standard. This would enable me to judge the extent to which people can be fooled by CGI.

I decided to keep the same structure of questionnaire containing three images from stock photography websites and my own image. Participants were asked to pick one image out of the four options that they believed to be computer-generated.

The results showed that out of the people who participated, a very high seventy five percent of them correctly chose photo B as the computer-generated image. The main reasons given for choosing this image were unrealistic lighting and the image having a “video game look”. This suggests that the image I created is not necessarily up to standard, in terms of realism. Only a quarter of participants were fooled into picking A, C or D, suggesting that the image used in the questionnaire needs to be improved. In order to achieve the same effect as the images of Alex Roman, CGI has to include a realistic lighting system. This highlights the gap that there can be between the standards of two different computer-generated images. By reworking and improving work that has been done in the past should enable me to create a very realistic image.

Artefact Two Feedback

My second artefact has given me results which are even more supportive of my research conclusion.

I decided to once again produce a questionnaire containing four images. Again, three of these were photographs, and the other was a computer-generated image by Alex Roman. Participants were asked to pick one image out of the four that they believed to be computer-generated and to give any reasons for their choice.

Results showed that out of the people who participated, a very low twenty percent of them correctly chose photo A as the computer-generated image. The main reasons given for choosing photo A were that the lighting and focus were not accurate, as well as some of the photo being blurred. In the image, Alex Roman has tried to give the illusion of filmed reality by adjusting the focus using depth of field. Eighty percent of participants were fooled into picking B, C or D, suggesting that computer-generated imagery can convince people by showing techniques used in photography. People who picked the wrong image gave reasons such as the textures not looking real and the images looking "cut-out". From this, it shows that Roman's image effectively mimics reality to fool the majority of people.

Client Project Update - Short Clips

Artefact One Feedback

My first artefact returned results which supported the conclusion of my research document.

I produced a questionnaire containing four images, from which the participants had to choose from. Three of these were photographs, and the other was a computer-generated image by Alex Roman. Participants were asked to pick one image out of the four that they believed to be computer-generated and to give any reasons for their choice.

Results showed that out of the people who participated, only thirty percent of them correctly chose photo D as the computer-generated image. The main reason given for choosing photo D was that there was some repetition used within the bookshelf. This is only really obvious when you take the time to look closely at the image. However, the fact that seventy percent of participants were fooled into picking A, B or C, suggests that computer-generated imagery can mimic realism very well. People who picked the wrong image gave reasons such as lack of detail and unrealistic light levels. From this, it is obvious to see that Alex Roman's image must have very accurate lighting and show a high level of detail.

Client Project Update - Single Frame Renders