Images in hover tags

[jackparmer]

I have links to the PNG and SVG molecular structure diagrams for each of the points in this 3d scatter plot (each point represents a molecule):

It would be awesome if images could be configured to show up in hover text.

2 screenshots of this feature in Spotfire for illustration:

[johnpvaillancourt]

I have the same requirement in a Plotly 3D scatterplot I'm working on in R, using plotly/htmlwidgets. Ideally, I'd like to be able to have an image attached directly to the hover box, for quick preview before opening a larger, external version for more in-depth inspection. Currently, I have images opening in separate windows by "plotly_click" on individual data points.

[chriddyp]

Community question about this here: #1323

[jackparmer]

Community question (R): #9223

If any companies are able to fund this feature, please reach out.

[emmanuelle]

Community question in https://community.plotly.com/t/how-to-create-customised-hovers-in-plotly/36815

[emmanuelle]

Mentioned in the forum https://community.plotly.com/t/graph-hover-feature/40421/2.

[jackparmer]

For visitors who are discovering this issue for the particular use case of displaying a molecular structure on hover, they may find these Dash molecule renderers useful in the meantime:

2d - https://dash.plotly.com/dash-bio/molecule2dviewer
3d - https://dash.plotly.com/dash-bio/molecule3dviewer
3d with ambient occlusion - https://dash.plotly.com/dash-bio/speck
3d based on NGL - plotly/dash-bio#496

[drscotthawley]

Four years and still open. :'-( I love the text annotation hover, & figured there's GOTTA be image annotation hover in a slick pro package like Plotly! ;-) Sadly I don't have molecular diagrams, just regular images.
Will keep a watch on @rreusser 's fork!

EDIT: Looks like Bokeh can do something along these lines -- even in Colab when used in concert with the nice package for Colab called kora. However the images must be hosted elsewhere -- bit of a wrinkle, that. For those interested, here is a demo notebook I made.

[zadamg]

Someday I will be able to put an image or picture into a plotly chart's hoverover.

It will be a great day. Today though, I am sad to see no progress thus far :*(.

[daviguima]

BUMP! That would be a great addition to my time-series analysis. Currently it is a pain to have several plots in parallel to inspect what is happening inside my simple px.line plot. But I know this is probably no trivial task at all, if it was, it would certainly be fixed already. Regardless of that, thanks for the great work you guys are doing behind the scenes in plotly (-:

[raphaeljolivet]

Hi, I needed the same, so I wrote a simple function to do it and posted in to this gist:
https://gist.github.com/raphaeljolivet/f076256e589f67c028a6bffaab279601

Example usage (assuming your images are stored locally) :

plot = xp.scatter(df, "nb_x", "nb_y")
template="<img src='./img/{id}.png'>"
interactive_plot(df, plot, template)

Output :

[drscotthawley]

@raphaeljolivet This is great.
And if the images are hosted on Colab, I wrote a routine to generate URLs for thumbnails of the images (requires the user to respond to a Google authentication prompt first):

https://github.com/drscotthawley/mrspuff/blob/master/mrspuff/scrape.py#L133

(!pip install mrspuff)

[jackparmer]

Images on hover is now finally available. Thank you everyone for your patience while we got this right and up to Plotly standards.

Huge thanks to @rreusser for getting the ball rolling.

Docs:
https://dash.plotly.com/dash-core-components/tooltip

Jupyter notebooks example 1:
https://colab.research.google.com/drive/1xUDTAK6NkzsLAYelCBMjpB09LjeI-faU?usp=sharing

Jupyter notebooks example 2:
https://colab.research.google.com/drive/1nSGnnlt_dwNFDNxCMIuDxk-AsujIdncP?usp=sharing

Screenshots!

If you like this feature, please give it a retweet here to help get the word out:
https://twitter.com/plotlygraphs/status/1445054301991211023

[drscotthawley]

@jackparmer This is great! Is it only available via Dash, or also via Graph Objects?
( e.g. https://plotly.com/python/graph-objects/tooltip 404's)

[jackparmer]

Only Dash

…

On Oct 4, 2021, at 2:04 PM, Scott H. Hawley ***@***.***> wrote:  @jackparmer This is great! Is it only available via Dash, or also via Graph Objects? ( e.g. https://plotly.com/python/graph-objects/tooltip 404's) — You are receiving this because you were mentioned. Reply to this email directly, view it on GitHub, or unsubscribe.

[Jmegan042]

Quick question about exporting this as an HTML - put simply: I haven't figured out how to do this. I can export the app using the tips from the plotly dash docs - and clicking the download button I create, but the resulting figure doesn't contain any of the hover information when I do.

I understand that this may be due to the way we update the app, but I was wondering if there was an idea of how to bypass this?

[Farshudsor]

Quick question about exporting this as an HTML - put simply: I haven't figured out how to do this. I can export the app using the tips from the plotly dash docs - and clicking the download button I create, but the resulting figure doesn't contain any of the hover information when I do.

I understand that this may be due to the way we update the app, but I was wondering if there was an idea of how to bypass this?

Also interested in figuring out how to export the interactive plot to HTML. as @Jmegan042 mentioned, the exported figures do not have any hover information.

[roel-heremans]

Thank you for the examples that show figures on hovering over. I am creating a sunburst plot and would like to show png figures when hovering over a sector of the plot. I was creating the following code already but when i hover over i do get a super long string to be displayed rather than my png figure.

The code i have so far is the following:

`def create_scr_sunburst_figure_hierarchy(sunburst_data, which):
res = []
for hierarchy, mape, mdape, mape_std, mdape_std in sunburst_data:
h_split = hierarchy.split("_")
res.append(h_split + [mape])

df = pd.DataFrame(res, columns=['Line', 'Group', 'Subgroup', 'mape'])

# Create a Sunburst figure
fig = px.sunburst(df, path=['Line', 'Group', 'Subgroup'], values='mape',
                  color='mape', color_continuous_scale='rdbu')

# Function to load the PNG file for a sector
def load_png_for_sector(hierarchy):
    # Define the filename for the PNG
    pre_fix = 'SCR_Regression (M. Chain)_'
    post_fix = '_Q0_scf_v2_neutral.png'
    pre_fix = 'fig'
    post_fix = '.png'
    filename = os.path.join('plots', 'SCR', 'Regression', 'all_pngs',
                            f'{pre_fix}{hierarchy}{post_fix}')
    # Check if the file exists
    if os.path.exists(filename):
        # Read the PNG file and convert it to base64
        with Image.open(filename) as img:
            img = img.resize((100, 100))  # Resize to an appropriate dimension
            with img as img_data:
                img_data = img_data.tobytes()
                encoded_image = base64.b64encode(img_data).decode()
            # Create an HTML image tag to display the PNG in the hover
            return f'<img src="data:image/png;base64,{encoded_image}" width="200" height="200">'  # Adjust width and height as needed
            # return f'<img src="data:image/png;base64,{encoded_image}">'
    else:
        return "No image available"

# Update the hovertemplate to include the image
fig.update_traces(hovertemplate='<b>%{label}</b><br>Parent: %{parent}<br>MAPE: %{value}<br>%{customdata}',
                  customdata=df.apply(lambda row: load_png_for_sector('_'.join(row[:3])), axis=1))

# Show the figure
fig.show()

# Save the figure as an HTML file
fig.write_html(os.path.join('plotly', f'sunburst_{which}.html'))

if name == 'main':

    sunburst_data = [
    ("Sector1_Subsector1_Category1", 0.1, 0.2, 0.3, 0.4),
    ("Sector1_Subsector2_Category1", 0.2, 0.3, 0.4, 0.5),
    # Add more data
    ]
    create_scr_sunburst_figure_hierarchy(sunburst_data, "example")`

When i hover over one of the sectors i do get a long string not a figure. ALthough when i save the information
from the return statement f'<img src="data:image/png;base64,{encoded_image}" width="200" height="200">' in a file and open that html file in a browser i do see the correct png.
Does anyone can help me out.

[yuxi-liu-wired]

Images on hover is now finally available. Thank you everyone for your patience while we got this right and up to Plotly standards.

Huge thanks to @rreusser for getting the ball rolling.

Docs: https://dash.plotly.com/dash-core-components/tooltip

Jupyter notebooks example 1: https://colab.research.google.com/drive/1xUDTAK6NkzsLAYelCBMjpB09LjeI-faU?usp=sharing

Jupyter notebooks example 2: https://colab.research.google.com/drive/1nSGnnlt_dwNFDNxCMIuDxk-AsujIdncP?usp=sharing

In case anyone gets an error, here's how you fix it:

Install these versions:

!pip install --q dash==2.18.0 jupyter-dash==0.4.2;

and for the MNIST example, do this line replacement

flattenend_images = np.array([i.flatten() for i in images])