Posts by bartneck

Exposure bracketing and rendering in AutoPano Pro

Posted by on Aug 24, 2018 in Design, Documentation, Featured | 0 comments

Exposure bracketing and rendering in AutoPano Pro

Taking 360 degrees panoramic photographs has become easy due to the arrival of dedicated 360 cameras, such as the Ricoh Theta V. But their 4K resolution is spread thin across the whole 360 degree viewing angle. There is still a need to panoramas with high resolution that can only be taken by stitching multiple photographs together. The GigaPan Epic Pro is a robotic camera head that tilts and pans your camera before triggering the shot. You can stitch hundreds even thousands of photos together into one GigaPixel panorama. I have uploaded a few on GigaPan’s website.

One problem when shooting GigaPixel panoramas is that some parts of your panorama will be very light due to the sun shining directly on them while other areas might be very dark, for example by simply being in the shadow. Finding a compromise exposure for your camera might be difficult and hence we are using bracketing to shoot multiple photos of the same area. Each of these photos will be several exposures apart from the others. For any shot you will have in fact three shots, each with a low, medium and high exposure. When you stitch the panorama together in AutoPano Giga, for example, you can use all three exposures. Autopano has several modes to select the best exposed pixel.

There are two settings that you can play with that will heavily influence how your final panorama will look like. You can set the Color Mode to None, Auto and HDR.

You can also set the Blending Presets to Simple, Anti Ghosting, Exposure Fusion and HDR Output.

I systematically varied these two settings to get an impression on what results the may produce.

Read More

Interview at UXPod with Gerry Gaffney

Posted by on Jul 28, 2018 in Press | 0 comments

Gerry Gaffney over at UXPod interviewed me about Robots, AI and Ethics. The transcript is available on his web page. Listen to our conversation below.

Read More

Interview with Jesse Mulligan on Radio NZ

Posted by on Jul 27, 2018 in Press | 0 comments

Jesse Mulligan interviewed me on our study Robots and Racism. It is good to see this topic is gaining some traction in the media.

 

Read More

Interview with Radio ABC about Robots and Racism

Posted by on Jul 20, 2018 in Press | 0 comments

Damien Carrick from ABC’s National Radio Show interviewed me about our Robots and Racism study. You can listen to the whole interview:

 

Read More

Visions of our android future

Posted by on Jul 10, 2018 in Culture, Featured | 0 comments

Visions of our android future

I started to play Detroit – Becoming Human and the start of the game introduces us to a world in which most of the work is done by androids. The designers paid attention to portraying every day life with androids. They show androids in the roles of care taker, cleaner, construction worker, but also as a companion. Many humans are without a job but still enjoy an android cleaning their mess. Of course this whole setup is a typical in-out-group setup. But if this is the future we are working towards then we will also have to address these questions. What if there is no more work left for humans? What if we like interacting with androids more than with other humans? Here are some of my highlights of everyday life in the game so far:

Android Parking

Detroit: Become Human

Android compartment in the back of the bus

The androids have to enter the public bus at the rear and are standing in a segregated compartment that is divided by a glass wall. This is of course a reference to Rosa Park who refused to give put her seat in the colored section of a bus for a white person in 1955.

Detroit: Become Human

An article on how autonomous cars make life and death decisions

The game features an article on how autonomous vehicles make life and death decisions by considering a wealth of background information to calculate the value of a person.

How machines make life and death decisions

Here is the text of the virtual news article:

When a driverless vehicle foresees an accident, the car’s computer makes life and death decisions – for example deciding which of two pedestrians to hit. But the exact process by which cars make these decisions is not very well understood.

Martin Forlong, of CrowneCars, tries to clarify: “In these situations, the car’s imaging system gathers data to determine, the pedestrian’s age, gender, life Expectancy, etc, in the blink of an eye.” This data is parsed through the public I record “to determine marital status, employment record, life expectancy and whether they have children.” The car then assigns a ‘value’ to each possible victim based on criteria like their contribution to society: “we put a premium on lives that will save other lives, like doctors and nurses.”

All this may sound very reasonable. But Felix Gamble, head of Anti-Automation League (AAL) says the system has no rights to make such judgements: “Somebody with a criminal record is not necessarily less I valuable to society than a doctor. That kind of information is irrelevant to the sanctity of human life.”

But Forlong dismisses such claims: “We want our cars to make the best possible i choices, and that means acting on the basis of all the information they can gather. The more, the better.”

Read More