REPORT: Input lag Part 2
The questions which need to be asked are as follows:
Is the stopwatch program precise enough?
In the course of the investigations for this test, it was necessary to program a new application which was capable of calculating and outputting an updated time display one thousand times per second. Some stopwatches which are now widespread are based on Flash or other programming languages which deliver sluggish 2D output or are necessarily linked with the screen refresh rate. Here, in the ideal case, a precision of just 16.67 ms for the time output arises, which corresponds to the time span for which a single frame is displayed at a refresh rate of 60 Hz.
How great is the lag of a CRT?
Thus, a close look at a CRT is first necessary in order to establish that there is indeed no or at least no relevant time delay in the signal output of a CRT. Until now this has been claimed without proof and is also used as a basis for the entire process.
Are the signal outputs on the outputs of a graphics card synchronised?
In the justifications for the accuracy of the photo method to date, the fact that the results are "similar" when the graphics card outputs are exchanged was used to conclude that there are no errors in the measurements. However, whether the signals are emitted synchronously, i.e. at the same time, has never been investigated or proven.
Here, of course, the question must be asked whether there are even manufacturers or model dependencies with regard to the graphics card used which could affect the measurements.
When does image content count as displayed?
In the course of this test, measurements and sample pictures are used to illustrate that due to the lack of standards or established measuring methods, it is difficult to establish from what point an image is considered "fully displayed" and whether this represents the decisive point for a measurement. A certain amount of time passes before old screen content is both considered locally and completely overwritten. Here, the switching speed of the individual transistors, or the response time of the panel used in the monitor, plays a role.
Errors in use of the camera
If you select long exposure times so that a CRT appears to be continuously illuminated in order to achieve "friendly" and readable images even for lenses that are weak in terms of light, it is often the case that no useful values can be read.
The reason is mainly that the refresh rate is not synchronised with the camera. Ideally, you would wish to create a moment photo which depicts the status precisely after the new refresh of the stopwatches. If you expose for a longer period, it can happen that for roughly half the exposure time, one or both of the stopwatches will be overwritten by new values.
The figures overlap and become difficult or impossible to read and, in the extreme case, even inaccurate if, for example, with certain figures and the display format of a seven-segment display turn several individual figures into another. Here, there is a very wide range of possibilities, which will be demonstrated in the pictures below using only the possible overlaps for the figure 4.
Possible overlapping of a seven-segment display with the number 4.
It can clearly be seen that there is just one overlap with the number four that is not a valid number and can therefore be recognised easily as an invalid value. The numbers eight and nine often arise as results and otherwise, it is sometimes impossible or at least difficult to see from such photos that you have become the victim of a mistake.
This effect can only be overcome using different number displays and displaying several stopwatches at the same time, allowing the refresh rate to be tracked.
This report focuses on the testing, evaluation and improvement of the measurement methods used to date. To this end, it was immediately necessary to develop a measurement method which is not subject to the limitations that have affected other methods to date, which is independent of the PC hardware used and which can be reproduced by third parties.
However, for precise measurement of the input lag, considerable technical outlay is necessary, which makes it practically impossible for the home user to check the results acquired in the course of this article at home.
After a program was developed which guarantees that precise figures are shown and displays the individual milliseconds precisely, the input lags of various monitors can be recorded using the photo method, as was the norm until now. These will have to be compared with the values obtained using the newly developed method at the end, taking into consideration all new knowledge acquired by then.
The image signals produced by the graphics card must be compared with each other. Here, mixed usage, i.e. one analogue output and one digital, is important, since CRTs are run on analogue, whilst LCDs should ideally be run digitally for the best image quality and also to avoid additional lags.
© 2002 - 2017 PRAD ProAdviser GmbH & Co. KG | All rights reserved! | Imprint | Privacy Notice
Registered trademarks and brand names are property of their respective owners.
PRAD assumes no liability for the content of external pages that we link to!
|General information: Our website is not partially displayed correctly with adblocker enabled!|