Portfolio thumbnail for FilterPixel

FilterPixel

FilterPixel is a productivity desktop application equipped with AI capabilities, built from the ground up to fit the needs of photographers & prosumers, allowing them to automate the repetetive & mundane tasks within their existing workflow, such as selection, editing, retouching and publishing the final delivery to their client.

Client

FilterPixel

My Role

Research, User Flows, Information Architecture, UX Writer, Wireframing, UI, Prototyping, Product Design

Problem

Ever since Adobe Lightroom revolutionised digital photography editing, the photography workflow has been a typically manual process. Photographers would shoot their images, import them to local storage, spend hours sifting through hundreds, if not thousands, of images, edit their selected photos and then find a way to deliver these to their clients. The process is often tedious and repetetive, costing photographers immense time.

Goal

FilterPixel's vision is to enhance the photographer's productivity. At the heart of FilterPixel is their in-house AI technology, with which they intend to automate part of the photographer's workflow. AI and Machine Learning within photography is still a fairly unexplored area, and so employing this technology without being intrusive will be key.

Insights from Discovery Phase

Hours are spent on the photo selection process, repeating steps

Constantly analysing faces for focus and whether eyes are open

Existing solutions for photo selection & organising are not intuitive

Finding the best photos out of duplicates is a chore

Available file sharing tools are not well suited to sharing photos with clients

The Solution

Given the large scope of the project, the solution was one that facilitated a multi tiered approach to enhancing productivity seamlessly with AI. From research it was clear that photographers were already set within the current structure of their workflow, so the solution would aim to greatly enhance the current experience without alienating it, whilst providing powerful new tools that feel fresh yet comfortable, allowing photographers to jump right in.

FEATURE ONE

AI INTEGRATION WITH TAGS AND FILTERS

With the AI enhanced Tag system, users able to quickly apply specific tags to photos and categorise them in place, whilst the AI applies Tags after analysing photos on import. During this analysis on import, FilterPixel's AI detects potential errors and applies relevant tags to images. Examples of AI generated tags include "Underexposed", "Group Photo", "Portrait Photo" and "Out of Focus", to name a few. The user then uses the Filters panel, to filter their view based on these tags, allowing them to easily see photos categorised under various Tags quickly.

Tags and Filter Panel In View

Tags and Filter Panel View

The Tag Panel works with the Filter Panel

Tag Panel working with Filter Panel

FEATURE TWO

AI ENHANCED FACES VIEW

In our discovery sessions, we decided to pivot the target audience for FilterPixel to event and wedding photographer's. Within the photography niche, these are the photographer's who will often rack up with 1000's of photographers from a single event, and who would benefit the most from a productivity enhancing software. With this in mind, the Faces view was designed to allow rapid navigation through photos, and allow quick analysis of the subjects within the image, without wasting time opening and closing individual images checking for the same repetetive mistakes; closed eyes, focus levels and exposure.

The Face Panel, docked to the left

The Face View

Multi-view for flexible visibility

Face Panel One Column ViewFace Panel Two Column View

Face Panel Explained

Face Panel Icons Explained

FEATURE THREE

MULTIPLE VIEWS WITH CUSTOM PANEL WORKFLOW

As FilterPixel would slot in to the photographer's existing workflow, it was imperative that the user experience and interface aligned with other softwares within the ecosystem. Photographers have come to expect certain features, so our intention was to take these expectations and push them beyond the scope of which other applications offered. Similar to Lightroom, Capture One and Photomechanic, FilterPixel has multiple views for photo selection and viewing. However, we implemented a unique panel based interface system that would give the user maximum customisation of their workflow.

Grid View

FilterPixel Grid View

Focus View

FilterPixel Single View

Split View

FilterPixel Split View

Tag View

FilterPixel Tag View

Fully customisable workflow

An iMac demonstrating FilterPixel Component View
A second iMac screen demonstrating FilterPixel Component View

My Process

Understanding the users

Prior to my employment, FilterPixel conducted a series of interviews with photographer's from across the globe. These findings were invaluable as they helped provide an understanding of the daily lives of different photographer's, their workflow, their schedules and most importantly, what common experiences and pain points they shared that could drive the vision of the app going forward.

You can view the interview details here. Interviews were conducted by FilterPixel over Zoom.

From the interviews it was revealed that:

Saving Time Icon Vector

Despite the nature of the role, photographers spend a significant amount of time on tasks outside of the actual photoshoot. This includes managing business operations, selecting and editing photos and delivery to the client.

Edit Icon Vector

Editing and Culling photos is the largest pain point commonly faced by photographers due to the repetetive and tedious nature of the role.

Automating Icon Vector

Photographers would gladly invite automation or a new tool into their existing workflow that eliminated these tasks if it worked seamlessly as part of the process.

Competitor Analysis

Considering FilterPixel would be entering an already crowded marketplace, we knew we had to look at what our competitors were doing and how they were doing it before making any further decisions on the product itself. I analysed several key competitors in the market in detail, you can find the analysis round-up below.

Click here to view the competitor analysis round-up.

From our competitor analysis we surmised a few things:

Plugin Icon Vector

It would prove to be a difficult task to infiltrate the market without being able to integrate with either the Lightroom or Capture One workflow seamlessly, given they are used by a wide share of photographers. We decided that we would develop FilterPixel as both an app and a plug-in.

AI Icon Vector

AI photo culling is a fine balancing act. We found users of Optyx were not particularly keen on the subjective nature of its culling, whilst Narrative Select did not provide enough information with its culling. To tackle this, we will be conducting thorough testing of our approach to AI culling and how the UI displays this.

Payment Model Icon Vector

Subscription models are not as popular as easy one time payments, however subscription models are healthier for long term growth and product delivery to the user.

User Persona

With the groundwork of research complete, I used this research to build the ideal user persona for FilterPixel; David.

FilterPixel User Persona Henry Hoult

Designing Features

As FilterPixel would need a variety of features, we brainstormed several ideas using user insights to understand ways in which we could provide solutions to consistent problems.

FilterPixel How Might We and POV Statements UX Task

How might we speed up the culling process so Henry can focus on the parts of his job that matter?

  • Use AI culling to handle the entire process from the onset of importing photos, such that suggestions for average to bad photos are automated by the software and presented to the user as either "Warnings" or "Rejected" images.
  • Provide a toggle to group photos that are similar.
  • Allow filtering through photos with speed by Tagging them and allowing users to customise Filter sets.

How might we provide a way for Henry to do all he needs to do in a single view?

  • Implement a customisable workspace that does not limit what is viewable in one area.
  • Place different features within flexible windows or "panels" that snap to the sides of the window.
  • Create a "Faces" panel that shows only the subjects faces in a zoomed in view, for quick viewing.

How might we deliver this information with AI to Henry?

  • Use a colour coded system to suggest what is "AI driven".
  • Provide information in the "Faces" panel that allows users to quickly see whether the subjects face is in focus or their eyes are closed.
  • Display quality metric scores for each photo, for a more detailed analysis.

Information Architecture


FilterPixel Information Architecture

User Flow

The aim of this product is to create a simplistic front-end process driven by a complex backend. During the brainstorming of this process, we came up with many various solutions and flows, but often found that these were too complicated. If the user was being forced to make too many macro decisions in the software, then the productivity aspect would fall apart. Therefore I decided to boil down the app into 3 macro decisions; one on launch, one after the first AI cull, and the final before exporting the photos to an Editing application.

Current Flow

FilterPixel User Journey

Goal Flow

FilterPixel Goal Journey

Wireframing

Given the already complex nature of south asian weddings, I felt that focusing on simplicity in design would be a firm anchor point for the app. Creating a rich design system that felt alive without being distracting or daunting would give the user the impression that planning a south asian wedding won't be a stressful experience, but a joy. I was especially mindful of the fact that competitor apps were filled to the brim with information on every screen, something I wished to avoid.

Low Fidelity Wireframes for FilterPixel

Validating

Once the mid-fidelity wireframes complete, I held six moderated usability tests, a mix of both remote and in-person. My goals were to:

  • Observe points of friction and areas of delight
  • Decide which options to make available in the tab menu based on user needs expectations

Some of the data I gained from the research and testing process can be seen below.

Validation Results from FilterPixel Wireframes

Improving the Review Panel

Improving the Review Panel for FilterPixel

User Interface

Once the mid-fidelity wireframes complete, I held six moderated usability tests, a mix of both remote and in-person. My goals were to:

  • Observe points of friction and areas of delight
  • Decide which options to make available in the tab menu based on user needs expectations

Some of the data I gained from the research and testing process can be seen below.

A showcase of the User Interface A showcase of the User Interface Icons

Retrospective

Reflection

Designing a competitive Photography application from the ground up was a genuine challenge, and yet it was also highly rewarding. As the team were working entirely remote, and user interviews were done with people all over the world, a large part of this application involved fascilitating this worldwide collaborative environment.

With FilterPixel, I contributed to the design of a product that solves a major problem within an Industry. As the engineers built the application, it became clear with every version iteration that this has the potential to be a market leader, as constant user interviews allowed both the developers and the design side to validate the product in build status.

Going forward, the team at FilterPixel is looking to launch to the market in Q1 2021 as a subscription based service.

Let's talk!

Mail Vector icon Linkedin vector icon