Coffee Extraction Photography: A Novel Way to Photograph Espresso
Seeing the bottom of an espresso shot has been obscured by the shot itself, until now.
I have long been interested in the bottom of the espresso shot. Ever since I got a bottomless portafilter, I had a thirst for more data. How could I better understand the espresso shot? I took measurements with a grid to break down the shot across the filter, but that was a difficult process to repeat every time. Around this time, I also started experimenting with the paper filter method (Paper Filter in the Filter or PFF), and I noticed something. After the shot, there was a negative image on the paper filter of the flow of the shot, and this could be valuable information that is otherwise invisible to the barista.
Typically, when you pull a shot even with a bottomless portafilter, you see a beautiful cone of coffee streaming into the cup. However, I found something different. That cone obscured the donut, and I discovered that from pressure pulsing. So how is one able to see after the shot starts?
The Paper Filter in the Filter (PFF) Method
Adding a paper filter into the espresso filter basket was popularized in May 2019. The technique involves adding a damp Aeropress paper filter cut to the size of the basket on the bottom, top, both, or something in the middle. This addition seems to improve taste and extraction. Putting a paper filter on the bottom of the filter allowed for higher extraction was due to using a finer grind than normal. The thought was that the grounds didn’t clog the filter, but I suspect the filter was causing channeling, and the paper filter lessened that channeling rather than channeling caused by too fine of a grind.
The second trick was to do a really long pre-infusion with just over 1 bar of pressure. Pre-infusion for most lever heads is between 5 to 20 seconds. Some have gone as high as 60 seconds, and I have found between 15 to 30 seconds is optimal depending on when the pre-infusion output starts to blond.
I started out testing PFF, but I was doing a Staccato shot. Through experimentation, I found the top filter to be unnecessary. For the bottom filter, the best place was between the fine and coarse layers or in the middle of the shot of staccato tamped shots. I suspect the fine layer was fusing together during pre-infusion which is why it wouldn’t clog the filter holes.
I started using PFF at the bottom of regular shots, and that’s when I really noticed an interesting pattern. The areas of the filter with slow water flow had high concentrations of coffee sitting there for a period of time, creating a dark spot. This can be seen on the bottom of a puck as well, but it isn’t as easy to spot because it is often only a little darker than the surrounding coffee.
Imaging with PFF
The main goal of this type of imaging was to understand whether I had good or bad shots. A good shot would have even distribution and few dark spots on this paper filter. A bad shot would have spots all around or clustered to one side. Spots in certain locations could be indicative that the filters hole variations were causing channeling.
I decided to run some experiments. One involved the staccato shot. I wanted to understand if there was channeling before the fine layer that impacted the channeling I saw when the coffee came out. This was part of my investigation to conclude that the way the water came through the shower head was causing the coffee to come out like a donut.
Here are some initial shots that showed some clear issues in the shot.
Let’s look at the fine layer of a staccato shot, and specifically putting a paper filter between the fine and coarse layers (staccato layering: fine on bottom, coarse in the middle, mid-range on top). I had to put two filters on top of each other because the one touching the coarse grounds was difficult to clean for viewing. The grounds wouldn’t come off, but for a double filter they would from one of the filters. I didn’t have to worry about the grounds from the fine layer because they fused together.
This case of Kompresso was particularly interesting because it aligned with my estimated flow from images of the filter. It was more extreme because I used a lot of a fines (<400um) grounds to make the shot so the channeling was more pronounced. Below is the predicted flow based on the filter analysis.
Let’s take a look at the pattern for one shot that showed pretty clearly the holes with predicted slower flow had coffee stains. This is caused by coffee getting to those parts of the filter, but as the shot progresses, channeling causes water flow to go elsewhere, and the coffee stays in that spot resulting in the stain.
Here is an overlay to help visualize:
Making Actual Measurements
We can take these images, these extraction shot images and take a look a flow. We can look at it the same way we can look at filter hole size. So I modified my code to bring in the PFF Extraction image and measure all the dots on the image.
In this case, flow was really high on one side of the filter but not the other suggesting an issue with the tamp being level. This doesn’t need a deeper analysis to see such an issue.
Imaging Mid-Puck Flow
One could also put the filters in the middle of the shot to help understand where the water flow is coming into the puck from the shower screen. For this shot, I put two white filters on top of each other, and I put those in a staccato espresso shot above the fine layer. There was some channelling in the shot, and on these, you can see the stains on the filter showing where water was coming through.
I was able to image the pattern showing the water coming from the sides of the filter, from four equally spaced points on the perimeter. The Kim Express has four evenly spaced holes for water to enter the chamber! Success! I used this information to help inform a decision to replace the shower screen to use an IMS shower screen.
When I started using paper filters, I did not expect such a result. It really helped me understand the flow of my shots, and it helped drive me to improve my distributions. The knowledge derived from photographing extraction helped in the development of the staccato tamp technique. It has also helped corroborate other evidence suggesting the hole analysis of the filters accurately predicted actual channeling.