Home‎ > ‎

Visualizing Flames


This page discusses techniques for imaging flames or taking pictures and video of flames and fire.  Using a Digitial Single Reflective Lens (DSLR) camera there are basically three setting you can change.  The shutter speed, the f-stop (aperture) setting, and the ISO.  The effect of changing these settings is shown using images of candles and Bunsen burner flames below.  The choice of lens is also important, but basically you want to be able to frame the shot you want to take.  


Photography books


When video taping flames the number of images per second is key in capturing move of fast moving flame fronts.  The use of video cameras is discussed below.


Flames can be visualized in a variety of ways:

Photography

Effect of Shutter Speed

Effect of shutter speed on visualizing a steady state flame.
The pictures taken in this section were taken using a Nikon D90.

On a steady state flame decreasing the shutter speed decreases the intensity of the image.
On a non steady state flame decreasing the shutter speed helps remove smearing caused by the movement of the flame.  Nominally one wants to use the shortest shutter speed possible while still resolving the image being sought.

Candle flame


This image shows the effect of changing the shutter speed when visualizing a candle flame in a normally lit room.  This is a good example of a luminous steady state diffusion flame.  ISO = 200.


Bunsen Burner Flame

This image shows the effect of changing the shutter speed when visualizing a Bunsen burner in a dark room.  This is a good example of a steady state premixed flame.   ISO = 200.


Shutter speed has a larger effect on non steady state flames.  The user must match the intensity requirements with the need to capture the flame location at a given time. The faster the shutter speed the lower the intensity or brightness of the image but the longer the shutter speed the more smearing of the image due to the movement of the flame during the time the shutter is open.

Effects of ISO

ISO refers to the sensitivity of the sensor.  The higher the ISO is set the more sensitive the sensor will be therefore giving you a brighter image but also a higher signal to noise ratio.  A high signal to noise ratio can cause random pixels to be the wrong color.  The below image shown a range of ISO settings used to take pictures of a candle flame with a set shutter speed and aperture (f-stop) setting.  It is shown that the intensity of the image becomes higher as the ISO is increased. 




Effects of Aperture (f-stop)

The size of the aperture has two effects, first it as the aperture becomes smaller the depth of field becomes longer.  The depth of field refers to the part of the picture which is in focus.  Therefore as you decrease the Aperture (increase the f-stop) more things separated in the z-axis will be in focus.  Below are images of a row of candles spaced 1 foot apart on axis (since the candle is off at and angle, the relative distance between the candles is a bit less than a foot) using various f-stops.  It is shown that the end candles become more in focus as the f-stop is increased.  But the second effect of the f-stop is to reduce the amount of light let into the camera as the aperture gets smaller.

Candles on lined up on table, 1 foot apart.

Effect of f-stop on candle flames 1 foot apart.


The effect of a large f-stop can be counteracted by increasing the ISO or the shutter speed.  Below is an image of having a large f-stop and increasing the ISO and changing the shutter speed.

Effect of ISO and shutter speed on candle flames 1 foot apart with a f-stop of 36.

The proper image setting will be a combination of shutter speed, ISO, and f-stop.



Propane Gas Burner

ISO = 200.

FRP C Wall Fire

FRP stands for a fiber reinforced polymer.
ISO = 200.

DSLR 

There are several benefits to using an DSLR. 

    - you can adjust the shutter speed, ISO, and f-stop setting as discussed above.

    - you can take multiple images without having to worry about film

    - you can use a wide variety of lenses and can rent lenses you do not want to buy

    -


D90

EOS 5D

    - Filters

    - Macro lens


Meta Data

Meta Data contains the information about the picture including when it was taken and the camera settings used.  It can also contain location data if a GPS unit was used with the camera. 

To read meta data of a picture in MATLAB use:
imageinfo(filename)
imfinfo(filename)

There are scripts that use these functions in the Matlab section of the site.





Infra Red - 87 filter (Tiffen)

The images in this section were taken with a Cannon EOS 5D and using an IR-87 filter to cut out the visible light.  Due to the low pass filter over the EOS 5D sensor the sutter speeds are generally 8 s to 30 s. 

Card board panel burning:



These images were taken using a shutter speed of 30 seconds.





Point and shoot camera

Sony Cybershot


Shadowgraph

Shadowgraphs are discussed in the Instrumentation/Diagnostics section.

Video

Digital Video Camera


Sony Handycam
records at 60 fps continuously or 260 fps for 3 seconds.

DSLR Video

The Nikon D90 records video at 30 fps for up to 5 min at high quality.  The focus will not automatically adjust while recording.

Cameras can also take pictures at a high enough rate to make a stop motion kind of video.  The following video of steel wool burning in a surface reaction was taken with a D90 at ~3fps.


MATLAB code to make a video like this can be found in the MATLAB section of this site.

High speed video

Casio

records at 300-1200 fps.  As the frame rate goes up the quality of the picture goes down.




Edge detection
    - best with a black background
    - free program made by NIH
    - can use matlab




It has been found that taking pictures at a rate of 1 fps makes it easier to match times to pictures without having to check the meta data for each picture.


Stochastic Techniques:

Sometimes it is not possible to predict when the picture needs to be taken.  For example the following pictures of lightning we taken using a Nikon d90 during a recent thunderstorm.  Approximately 1000 images were taken in a 15 minute period.  Out of these, three images turned out to be good.  The d90 can take images at a rate of 4.5 frames per second (fps) until the internal buffer fills and the camera starts to write to the card (this takes about 4 sec using a normal .jpg image) and then the camera continues to capture images at ~2 fps.  This is a function on how fast the camera can write to the memory card.  The better the quality images the faster the internal buffer will fill up and the longer it will take to write to the memory card.  This type of technique has been used for experiments such as portable gas can detonations where it is unknown if or when the flame will propagate up the nozzle and into the main reservoir.



It is difficult to know exactly how fast the camera is taking pictures once the buffer fills up because it is a function of the memory card type and the size of the picture you are taking.  One way to estimate the number of pictures per second is to take pictures of a running stop watch and look at the change in time.  This is only an approximate result but will get you in the ball park.  The images below show 2 fps being recorded, but they vary by ~10-20 hundredths of a second over time.


Unfortunately many DSLR camera will only time stamp the data to the second.  There is an output called subsec in the meta data which is not shown with the basic time stamp.  One way to find this information in the meta data using MATLAB is linked here.  Unfortunately the Nikon d90 does not report a subsec value.

Focusing camera:


Images like the above lightening are difficult for the camera to auto focus.  This difficulty is also found when attempting to take pictures of flames in a dark room.  To overcome this problem I use a target to focus the camera before the experiment and then set the camera to manual focus preventing the focus from changing.  It is important to remember to refocus if the camera is moved or the zoom is changed.






Organizing Images from Experiments:


Once a series of tests is completed, it can be difficult to remember which images go with which test conditions.  I like to use pictures of pieces of paper with the test number on them to keep the pictures in order.  When using long exposure times in the dark this can have some interesting effects if the piece of paper is removed before the shutter closes.  This image was taken using a 1 second exposure time, with the room dark and using a flashlight to view to paper.  You can see the flame behind the piece of paper and can thus verify that the following images go with that data set.


It is important to write what the conditions for each test are in a laboratory notebook.  More information can also be included in the seperation image as-well.  Though this tames more effort than a simple numbering system.  I generally use a set of sticky notes numbered 1-10 and then record the conditions each number corresponds to in my notebook.


Long Experiments:
For long duration experiments, such as smouldering, making a stop motion video is a reasonable way to capture the process.  An example of a stop motion video of a sunset with images taken every 10 seconds and played back at 20fps is shown below.   Tips for making stop motion videos can be found in the Making video tips section.



To rename large numbers of files without loosing the order in windows select the entire set of images, right click and select rename on the first image, rename something like image(10001).jpg (or whatever file type).  If you use image(00001) it will not hold the zeros in the sequence.  This is important to do when making videos using programs like the matlab script located on this site because the numbering will become messed up for every order of magnitude.  Aka the numbers will go something like 99, 10, 100...

Mounting Cameras


Typical camera mounts use a 1/4 inch standard thread.  When making repeat experiments it is important to have a stable place to mount the camera.  Standard tripod will work but they are difficult to place repeatedly and can be easily kicked over.  The most optimum option is to use an optical table but these are fairly expensive.  I use an extruded aluminum frame to mount my D90 for static experiments.  Knowing that the counting screw is 1/4 inch lets you use any mounting scheme that you want.  








Comments